Docstoc

IPO underpricing and long term underperformance

Document Sample
IPO underpricing and long term underperformance Powered By Docstoc
					         IPO underpricing and long-term underperformance

                                          Francisco Santos
                                            Stanford GSB
                                       <Job Market Paper>

                                         November 30, 2010



                                               Abstract

          This paper …lls in a gap in the IPO literature by documenting a close connection
      between IPO underpricing and the long-term underperformance of IPOs. Firms going
      public in periods of low underpricing do not underperform in the long run, while …rms
      going public in high underpricing periods do. Furthermore, IPOs in later stages of high
      underpricing periods underperform even relative to their o¤er prices, which suggests that
      many of the most "underpriced" IPOs are in fact priced above fundamental value. This
      result cannot be explained by di¤erences in risk, and it is unlikely to be driven by a
      peso problem. I also …nd that …rms going public in later stages of high underpricing
      periods display worse operating performance and pro…tability, lower asset growth, lower
      investment rates and higher cash holdings. Finally, I provide evidence that investor
      sentiment is stronger in high-underpricing periods. These results are consistent with a
      story in which low quality …rms, in periods in which the average underpricing in the
      market is high, try to exploit investors’sentiment by going public.




    I am extremelly grateful to my dissertation committee: Dirk Jenter (chair), Francisco Perez-Gonzalez,
and Ilan Kremer for many insightful comments and guidance. I am also grateful to Anat Admati, John
Beshears, Jules van Binsbergen, Darrel Du¢ e, Liran Einav, Arthur Korteweg, Charles Lee, Andrey Malenko,
Nadya Malenko, Paul P‡   eiderer, Ilya Strebulaev, Yanruo Wang and Je¤rey Zwiebel for helpful comments
and discussions. I am responsible for all remaining errors. Address for correspondence: Graduate School of
Business, Stanford Business, 518 Memorial Way, Stanford, CA, 94305. Email: fjsantos@stanford.edu


                                                    1
                      IPO underpricing and long-term underperformance


1    Introduction


                                                               s
The initial public o¤ering (IPO) is an important step in a …rm’ life and a central issue in

corporate …nance. As such, the topic has received widespread attention in the literature.

Enhancing the attention given to IPOs is the fact that IPOs are associated with some inter-

esting empirical patterns. Two of the most important ones, also called “anomalies” of the

IPO market, are positive …rst day returns (so-called underpricing) and long-run underperfor-

mance. The underpricing of IPOs is one of the most studied anomalies in …nancial economics

(McDonald and Fisher, 1991; Logue, 1973; Ibbotson, 1975; Ritter, 1984; Ibbotson, Sindelar

and Ritter, 1988; Ritter, 1991; Ritter and Welch, 2002; Ritter and Loughran, 2002; among

others), and this attention is driven in part by the magnitude of the phenomenom - for the

last 40 years, the average …rst-day return is around 20%. The long-run underperformance

describes the observation that IPO …rms, over the 1 to 5 year period, subsequent to the

IPO, tend to underperform relative to the market and relative to public …rms with similar

characteristics (Ritter, 1991; Loughran and Ritter, 1995; among others). Although extensive

research has been done trying to explain either IPO underpricing or the long-term underper-

formance of IPOs, the two issues are usually not analyzed jointly as most papers address one

issue without taking into account the existence of the other1 .

    The objective of this paper is to examine the connection between IPO underpricing and

long-term underperformance. More speci…cally, I analyze whether …rm-speci…c underpricing

or average underpricing in the market, de…ned as the average underpricing for all recent

o¤erings, determines the degree of subsequent long-term underperformance. My central …nd-

ing is the existence of a close relation between periods of high/low …rst day returns in the

market and subsequent long-term performance. Five-year wealth relatives comparing …rms

going public in low-underpricing or in the beginning of high-underpricing periods with public

…rms matched on book value, market value, industry and operating performance are not

signi…cantly di¤erent from one. Wealth relatives for …rms going public in the late stages of

high-underpricing periods are well below one. This implies that the long-run underperfor-

mance of IPO …rms documented by Ritter (1991) and Loughran and Ritter (1995), among
   1
     Notable exceptions are Ritter(1991), Krigman, Shaw and Womack (1999), and Ljungqvist, Nanda and
Singh (2006).



                                                 2
                     IPO underpricing and long-term underperformance


others, is caused by …rms going public in the late stages of high-underpricing periods. More-

over, assuming that the long-run price proxies for fundamental value, I …nd that …rms going

public in the late stages of high-underpricing periods underperform even relative to their

o¤er price. This result is di¢ cult to reconcile with the view that IPO underpricing is a

discount to fundamental value. Instead, it suggests that IPO prices, in these late stages of

high-underpricing periods, include a premium over fundamental value.

   Having established a strong link between average IPO underpricing in the market and

long-term performance, I subsequently examine the underlying forces driving such correlation.

   First, I show that di¤erences in systematic risk cannot explain the relation between IPO

underpricing and long-term underperformance. For risk to explain long-term underperfor-

mance, …rms going public in high-underpricing periods would have to be safer. I test this

explanation by looking at earnings volatility, delisting rates, total return volatility, and the

betas of …rms going public. There is no evidence that IPOs in periods of high underpricing are

safer than IPOs in low-underpricing periods. I also do not …nd that …rm risk decreases within

high-underpricing periods, which would be needed to explain why the long-term underperfor-

mance is caused by …rms going public in the end of high-underpricing periods. Furthermore,

and in contrast to the risk-based story, I …nd that large underpricing is correlated with sig-

ni…cantly weaker (not stronger) …rms. I show that …rms going public in high-underpricing

environments, subsequent to the IPO, exhibit striking di¤erences in terms of operating perfor-

mance. Speci…cally, these …rms exhibit worse ROA and pro…tability, invest less, display lower

asset growth, and hold more cash than …rms that undertake their IPOs in low-underpricing

periods.

   I also provide suggestive evidence that the relation between IPO underpricing and long-

term underperformance is not caused by a peso problem. For growth opportunities to explain

the relation between underpricing and long-run performance, we would need that …rms in

a high underpricing environment are, on average, worse, but also more likely to become

the next Microsoft, i.e. more likely to yield extremely high returns in the long-run. I test

this explanation by looking at the fraction of …rms going public in low-, neutral-, and high-

underpricing periods that earn extremely high returns, de…ned as more than 500 or 1000%,

in the …ve-year period after the IPO. I do not …nd that IPOs undertaken in high-underpricing

periods are more likely to deliver extremely high returns. Moreover, I …nd that the probability

                                               3
                     IPO underpricing and long-term underperformance


of yielding extremely high returns is the lowest in the late stages of high-underpricing periods,

when long-term underperformance is more severe. This is inconsistent with the view that

investors are paying high prices in high-underpricing periods hoping to discover the next

Microsoft.

   Next, I provide evidence that investor sentiment is stronger in high-underpricing periods

than in low-underpricing periods. First, I …nd that …rms going public in high-underpricing

periods experience a higher number of stock trades and higher turnover in their …rst day of

trading. This is consistent with the idea that over-optimistic retail investors are driving the

stock price above fundamental value on the …rst day of trading. Second, I …nd that three

widely-used measures of sentiment - the dividend premium, the percentage of equity issues

in total issues, and the University of Michigan Consumer Con…dence Index - indicate that

investor sentiment is stronger in high-underpricing periods. Third, I show that investors re-

act more positively to news in periods of high underpricing. Estimating investors’response

to earnings announcements, I …nd that cumulative abnormal returns on the day of the an-

nouncement and the next day are on average …fty basis points higher in periods of high

underpricing in the IPO market. This e¤ect is more pronounced for extreme positive news.

   These result are consistent with a story where some …rms go public to take advantage

of overvalued equity due to the presence of overly-optimistic investors. In periods of high

underpricing in the market, when IPO prices seem to include a premium over fundamental

value, …rms with and without investment opportunities have an incentive to go public. In

low-underpricing periods, when IPO prices do not include a premium over fundamental value,

only …rms that need …nancing for positive NPV projects have an incentive to go public. Hence,

post-IPO performance is di¤erent for …rms going public in high- and low-underpricing periods.

   Finally, looking at the magnitude of earnings surprises and the number of earnings es-

timates by analysts, I provide suggestive evidence that periods of high underpricing are

correlated with periods of high information asymmetry. As I will discuss in Section 5, this re-

sult may explain why …rms, although raising IPO prices above fundamental value, are willing

to leave money on the table in high-underpricing periods.

   This paper contributes to the IPO literature by establishing a close relation between

average underpricing in the market at the time of the IPO and subsequent long-term un-

derperformance. My results identify a need for explanations that jointly address both IPO

                                               4
                            IPO underpricing and long-term underperformance


underpricing and long-term underperformance. Moreover, my results are inconsistent with

the standard view that underpricing represents a discount to fundamental value (Rock, 1986;

Welch, 1992; Benveniste and Spindt, 1989; Benveniste and Wilhelm, 1990; and Spat and

Sristava, 1991) or a costly instrument that …rms use to signal quality (Welch, 1989; Allan

and Faulhaber, 1989; Chemmanur, 1993). Instead, periods of high underpricing appear to

be periods in which …rms are able to raise their IPO prices above fundamental value. Note

that this …nding o¤ers a simple explanation for the puzzle why so many …rms appear keen

on going public in periods of high underpricing. Firms leave money on the table but, at

the same time, they are getting more than what they are worth2 . Finally, I …nd that there

are, post-IPO, substantial di¤erences in observable operating performance for …rms that go

public as a function of the average underpricing in the market. To my knowledge, this is a

novel result. Theories trying to explain IPO underpricing and long-term underperformance

should also be able to explain di¤erences in operating performance.

                                                               y
        The outline of the paper is as follows: Section 2 brie‡ summarizes the prior literature.

Section 3 describes the data, the return benchmarks, and the procedures used to identify

high- and low-underpricing periods. Section 4 presents the empirical results. I analyze the

relation between IPO underpricing and long-term underperformance; …rm risk; the likelihood

of extremely high returns as a function of initial underpricing; the quality of IPO …rms;

and the relation between periods of high average underpricing in the market and investor

sentiment. Section 5 discusses potential explanations for why …rms may be willing to leave

money on the table even if they are extracting a premium from sentiment investors. Section

6 concludes the paper.




2        Literature Review


Empirical evidence on the relation between underpricing and long-run perfor-

mance

        This paper examines the relation between the average level of underpricing in the market

and the long-run underperformance of IPO …rms. Two prior studies have examined the
    2
        I will discuss several explanations for why …rms may be willing to leave money on the table in section 5.



                                                         5
                     IPO underpricing and long-term underperformance


connection between …rm-speci…c underpricing and long-term underperformance.

   Ritter (1991) documents that …rm-speci…c underpricing and long-run performance are

negatively correlated. Using a sample of IPOs for the period 1975-84, Ritter compares after-

market returns for quintiles of industry adjusted initial returns. He shows that …rms with

high underpricing have the worst aftermarket performance.

   Krigman, Shaw and Womack (1999) also look at the relation between …rm-speci…c un-

derpricing and long-term underperformance. They partition a sample of IPOs for the period

1988-95 by …rst-day returns and show that only the most extreme …rst day returns predict fu-

ture performance. They …nd that IPOs with underpricing above 70% and IPOs with negative

…rst day return underperform in the long run.

   I examine both …rm-speci…c and average underpricing in the market as a predictor of

long-run underperformance. I …nd that the average underpricing in the market determines

the degree of subsequent long-term underperformance. Firm-speci…c underpricing cannot

explain long-run performance once the market e¤ect is taken into consideration.


   IPO underpricing

   A sizeable literature provides evidence that IPO underpricing is a persistent phenomenon

in the IPO market - see Reilly (1973); Ibbotson (1975), Ritter and Welch (2002), Loughran

and Ritter (2002), among others.

   In response to the empirical evidence, the literature has developed a variety of theories

to explain underpricing. Most theories of underpricing rely on some type of asymmetric

information between issuers, underwriters and investors.

   If the issuer is more informed than investors, the money left on the table through un-

derpricing can work as a signal of higher quality (Welch, 1989; Allan and Faulhaber, 1989;

Chemmanur, 1993). Alternatively, in a model where some investors are more informed than

…rms, for example about market conditions, …rms provide a discount either to guarantee that

uninformed investors break even (Rock, 1986) or to induce informed investors to share infor-

mation (Benveniste and Spindt, 1989; Benveniste and Wilhelm, 1990; Spat and Srivastava,

1991; Welch,1992).

   Another set of theories tries to look at IPO allocations as the main driver for underpricing

- Benveniste and Spindt (1989), Booth and Chua (1996), Brennan and Franks (1997), Mello


                                              6
                    IPO underpricing and long-term underperformance


and Parsons (1998), Stoughton and Zechner (1998) Sherman (2000), Sherman and Titman

(2002), and Loughran and Ritter (2002) are examples. Underpriced o¤erings create excess

demand which allows discretion by …rms on whom to allocate shares. Implicit, is the argument

that some shareholders are more desirable than others.


   Long-run underperformance of IPO …rms

   Another large literature provides evidence that IPO …rms tend to underperform in a 1-5

year period subsequent to the IPO. For example, Ritter (1991), and Loughran and Ritter

(1995) report a 16 percent buy-and-hold return for IPO …rms while their comparable size-

matched …rms earn 66 percent over the …ve year period. Loughran and Ritter interpret their

evidence as investors being too optimistic about the prospects of …rms issuing equity for the

…rst time. Moreover, Loughran, Ritter and Rydqvist (1994) claim that …rms time their IPOs

to coincide with periods of excessive optimism, consistent with …ndings in Lee, Shleifer, and

Thaler (1991) that more companies go public when investor sentiment is high.

   Another contribution of my paper is that I look at buy-and-hold returns including and

excluding …rst day returns. Including the …rst day return means that the long term per-

formance is measured from the IPO price, while excluding the …rst-day return means that

performance is measured from the …rst day closing price. Previous papers examine long term

underperformance starting from the closing price at the …rst day of trading. Observing long-

term underperformance excluding the …rst-day return is not enough to claim that …rms take

advantage of excessive optimism. However, long-term underperformance including the …rst

day return implies that IPO prices are above fundamental value which would be consistent

with the claim that IPO prices include a sentiment premium.

   It is still unclear how abnormally poor post-IPO performance is. Long-run returns, even

if extremely low, are su¢ ciently noisy to make statistical inference di¢ cult. As Welch and

Ritter (2002) point out in Brav (2000) it can require an abnormal return of -40 percent to

reject the hypothesis of underperformance. However, the evidence strongly suggests that

IPOs and …rms with similar characteristics exhibited poor performance in times in which the

overall market performs exceptionally well.

   Miller (1977) tries to explain IPO long-term underperformance with short-sale constraints

on IPOs and heterogeneous beliefs about fundamental values. Immediately after the IPO, the


                                              7
                     IPO underpricing and long-term underperformance


marginal investor is the most optimistic among all investors but as time passes, di¤erences in

                                             s
valuations decrease and the marginal investor’ valuation converges to the mean valuation.

Hence, in the long run IPO …rms underperform.

   Most common explanations for the long run underperformance rely on some type of

overcon…dence - Teoh, Welch, and Wong (1998), Heaton (2001), Bernardo and Welch (2001),

Daniel, Hirshleifer, and Subramanyam (1998). The main idea is that investors in the short-

run overshoot fundamental value and in the long run prices revert to the correct level.

   To summarize, the prior literature in underpricing and long-term underperformance sug-

gests that …rms go public at a discount to fundamental value which leads to positive …rst day

returns; investors overshoot fundamental value on the …rst day of trading, and in the long

run, prices revert to their fundamental value.


   Connecting underpricing and long-term underperformance

   The empirical results presented in this paper are in line with the theoretical work of

Ljungqvist, Nanda and Singh (2006). Di¤erent from most of the IPO literature, which

usually addresses each issue separately, they develop a model of IPO pricing that connects

underpricing and long-term underperformance. A major assumption of their model is that, on

occasion, some investors are "irrationally exuberant" about IPOs. In a setting where shorting

IPO shares is di¢ cult or too costly, the presence of sentiment investors implies that some

periods should display long-term underperformance excluding the …rst day return. Moreover,

in times where these sentiment investors are likely to show up, …rms raise IPO prices in order

to exploit them. This is consistent with my result that in certain periods we observe long-term

underperformance even relative to the IPO price. Ljungqvist, Nanda and Singh (2006) model

also tries to explain why, although …rms raise IPO prices above fundamental value, they do

not fully exploit sentiment investors. In their model, high underpricing is compensation given

by …rms to intermediaries for bearing the risk of carrying overvalued IPO shares in inventory.




                                                 8
                     IPO underpricing and long-term underperformance


3     Data, IPO cycle and Benchmark


3.1   Data

The initial sample includes all US IPOs completed between January 1, 1973 and December

                                          s
31, 2008 as reported by Thomson Financial’ Securities Data Company (SDC) database. Unit

o¤ers, spin-o¤s, closed-end funds, IPOs with an o¤er price below $1.00, Real Estate Invest-

ments Trusts (REITs), American Deposit Receipts (ADRs) and …nancial …rms as described

by SDC are excluded from the sample. From SDC I obtain the date of issue, the dollar value

of proceeds raised and the …rst day return.

    I obtain trading price histories from the Center for Research in Security Prices (CRSP)

database. I exclude from the sample observations for which no match is found on CRSP based

on the Cusip reported by SDC. I also exclude observations for which mismatches between

CRSP and SDC data make it di¢ cult to identify the correct IPO date. In particular, I

exclude cases for which the trading price history on CRSP starts on a date prior or four days

after the reported IPO date by SDC. This yields a sample size of 6,256 IPOs for which there

is information on …rst-day underpricing.

    Table 1 provides information on IPO activity in the U.S. from 1973 to 2008. I report

the number of IPOs, average …rst day returns, average and aggregate proceeds, and average

and aggregate money left on the table per year. Proceeds are de…ned as number of shares

issued times IPO price; money left on the table is de…ned as proceeds times …rst day return.

As documented by the prior literature, we see that the number of IPOs per year is highly

volatile. In 1975, only 5 …rms undertook an IPO, while 599 …rms decided to go public in

1996. In terms of average …rst day returns, not a single year from 1973 to 2008 exhibits a

non-positive average …rst day return. As with the number of IPOs, the yearly time series

of average …rst day return displays high volatility, with values ranging from 1.9% in 1974 to

78.6% in 1999. Figure 1 displays the number of IPOs per month and the monthly average

underpricing; underpricing and volume volatility is clearly visible as well. The amount of

money raised per IPO is increasing overtime. For the entire sample period, the average IPO

raises 72.5 million dollars, but, this amount doubles for the 2000 to 2008 sub period. Overall,

this sample of IPOs was able to raise 496 billion dollars, but left 122 billion dollars on the



                                              9
                        IPO underpricing and long-term underperformance


table through underpricing. In 1999 and 2000 alone, the money left on the table amounted

to 67 billion dollars as a result of high underpricing and high volume.

   The magnitude of money left on the table is puzzling for at least two reasons. First, if

underpricing is a discount on fundamental values, why do we observe higher IPO volume when

this discount increases? Second, according to asymmetric information models, high levels of

underpricing are caused by severe information asymmetries between …rms, underwriters, and

investors. Hence, any strategy that helps decrease information asymmetries would lower

underpricing. For example, bundling IPOs would decrease the level of uncertainty about the

average …rm. However, we do not observe such strategies being implemented even when the

money left on the table amounts to 67 billion dollars over two years.


3.2   High- and low-underpricing periods

The primary goal of this paper is to study the relation between the level of IPO under-

pricing in the market, measured as monthly average underpricing, and subsequent long-run

performance. Motivating this choice is the fact information asymmetries between …rms, un-

derwriters, and investors can explain underpricing but not long-term underperformance. It

might then be the case that the long-term underperformance is caused by sentiment investors

driving up prices in the …rst-trading day. If …rms do not fully exploit these sentiment in-

vestors, we would observe high …rst day returns in the IPO market. Hence, it seems reasonable

that average underpricing in the market correlates with the presence of optimistic investors

more powerfully than …rm-speci…c underpricing.

   The …rst step is to de…ne if, at a given moment, the IPO market is displaying low or high

levels of underpricing. In line with Helwege and Liang (2002), I divide the IPO cycle into

three phases: low, neutral and high underpricing.

   I calculate a time series of the average monthly underpricing from 1973 to 2008. For

each month I compute the distribution of the previous 120 monthly average underpricing

measures. A month is considered high underpricing if its underpricing is higher than the

75th percentile; low if its underpricing is below the 25th percentile; all remaining months are

classi…ed as neutral.

   It might be that one single month of high underpricing is driven by outliers and does

not re‡ect the state of the IPO market. Hence, using low, neutral, and high months I

                                              10
                         IPO underpricing and long-term underperformance


de…ne low, high, and neutral periods. A high-underpricing period consists of three or more

consecutive high-underpricing months. Low-underpricing periods are de…ned analogously,

while all other periods still unde…ned are considered neutral-underpricing periods. In order to

capture potential dynamics within high-underpricing periods, I subdivide high-underpricing

periods into beginning, middle, and end. I do this in a simple manner by dividing each period

into three parts (and giving the middle part the biggest share when the number of months is

not divisible by three).

      The prior literature subdivides the IPO market into “hot” and “cold” periods based on

IPO volume, not underpricing3 . To compare results I perform a similar classi…cation. For

each month I compute the distribution of the number of IPOs per month during the last

120 months. A month is considered high volume if the number of IPOs in that month is

higher than the 75th percentile; low if the number of IPOs is below the 25th percentile; all

remaining months are de…ned as neutral. High-, low-, and neutral-volume periods are de…ned

analogously to what is done under the underpricing de…nition.

      Figure 2 shows that "hot" periods in terms of underpricing and volume do not coincide.

For example, in 1978 we see high underpricing but low volume, while in 1986 and 1987 we

observe high volume but low …rst day returns. In the analysis section below, we will …nd that

the two classi…cations of the IPO market have very di¤erent implications for the subsequent

performance of IPO …rms.

      Table 2 provides information on the number of IPOs and the average …rst day return by

the state of the IPO market. Panel 1 uses the classi…cation into months while Panel 2 uses the

division into longer periods. 688 …rms go public in low-underpricing periods with an average

…rst day return of 6.3%. In high-underpricing periods, the number of …rms going public

more than doubles and …rst day returns jumps to 41.6%. Looking at the volume de…nition,

there are no signi…cant di¤erences in average underpricing between high- and neutral-volume

periods. Within high-underpricing periods we do not observe signi…cant variance or any

clear trend in the level of …rst-day returns. However, the number of IPOs almost doubles

from the beginning to the end of high-underpricing periods. This increase in volume within

high-underpricing periods is puzzling if higher underpricing represents a large discount to

fundamental value, as suggested by prior literature. However, the increase in volume in high-
  3
      Notable exceptions are Ritter(1984), and Helwege and Liang (2002), among others.


                                                     11
                     IPO underpricing and long-term underperformance


underpricing periods makes sense if high underpricing coincides with IPO prices that are set

above fundamental value.



3.3   Benchmarking returns


A proper benchmark is fundamental to analyze IPO …rm performance. However, such bench-

marking is di¢ cult for …rms that just went public due to the lack of information about …rm

characteristics prior do the date of the IPO. In this paper, I follow the literature and match

IPO …rms with public …rms with similar characteristics at the time of the IPO. The criteria

used to match IPO …rms with public …rms is based on Lie (2001) and uses book value of

assets, market size, return on assets (ROA), and industry.

                                                       s
   For each IPO …rm, I retrieve from Compustat the …rm’ book value, market size, ROA,

and SIC code at the end of the IPO year. Then, for each IPO …rm, I try to obtain control

…rms from the universe of …rms on Compustat. I want a matching procedure that is precise

but also does not result in a considerable loss of observations. To satisfy these two goals, the

matching process implements sequential sets of criteria. Speci…cally, I de…ne the following

three sets: I - same three-digit SIC code, and market size, ROA, and book value of assets

between 70% and 130% of the IPO …rm values; II - same two-digit SIC code, and market size,

ROA, and book value of assets between 70% and 130% of the IPO …rm values; III - market

size, ROA, and book value of assets between 70% and 130% of the IPO …rm values. I do not

allow a control …rm to be chosen if its IPO was less than …ve years prior to the IPO of the

event …rm as I do not want IPO …rms controlling for IPO …rms. I apply criteria I and choose

the …ve …rms closest to the IPO …rm in terms of market size. If criteria I does not yield …ve

control …rms, then I apply criteria II. If I and II are still not enough I apply criteria III. In

the end, for each IPO …rm I obtain a portfolio of …ve control …rms as benchmark. Choosing

only the closest …rm as the benchmark, or using book-to-market instead of market size as

the factor to decide closeness to the IPO …rm, does not change the results presented below.




                                               12
                     IPO underpricing and long-term underperformance


4       Empirical Analysis


4.1     Underpricing and long term underperformance

Following Ritter (1991), Loughran and Ritter (1995) among other, the strategy implemented

to test for long run underperformance is to compare buy-and-hold returns for IPO …rms and

control …rms matched on book value, market value, industry, and operating performance.

    I compute …ve-year buy-and-hold returns using daily and monthly returns from CRSP. I

demand that IPO …rms have at least one year of return history after the IPO but if, after one

year, information is missing for an IPO stock I do not drop that …rm. For the benchmark

…rms, I only include returns for the period for which data is non-missing for the IPO …rm.

For example, if an IPO …rm has only four years of returns I use those four years for the IPO

…rm and benchmark …rms. This procedure mitigates sample selection bias by not dropping

…rms with limited return histories.

    Finally, for each …rm I compute 5y CAR1d and 5y CAR0d which represent, respectively,

…ve year cumulative return excluding and including …rst day return.

    Most studies of IPO long term underperformance exclude the …rst day return, reasoning

that the price at the end of the …rst trading day is a better proxy for …rm value than the IPO

price. I follow a more general approach by computing buy-and-hold returns both relative to

the IPO price and relative to the price at the end of the …rst trading day. Di¤erences between

these two measures should help identify the correct interpretation of IPO underpricing. If

underpricing is to be interpreted as a discount to fundamental value, then one should not

observe long-term underperformance relative to the IPO price. If, however, …rms underper-

form even relative to the IPO price, then underpricing cannot be interpreted as a discount

to fundamental value as it seems that IPO prices are set above fundamental value.

    The next step is to analyze the relation between long term underperformance, measured by

5y-CARs, and underpricing at the time of the IPO. For that purpose, I run several regressions

of 5y    CAR1d on di¤erent de…nitions of underpricing, testing for a relation at the …rm and

market level.

    Table 3 presents the …rst important result of this paper. There is a strong negative

relation between long-term performance of IPO …rms and the average IPO underpricing at



                                             13
                     IPO underpricing and long-term underperformance


the market level. Moreover, this long-run underperformance is more severe for …rms going

public after several months of high underpricing in the market.

   Column (1) shows a negative relation between …rm speci…c underpricing and subsequent

long-term performance. A one percentage point increase in initial underpricing leads to a

statistical signi…cant decline of 43 basis points (bps) in the 5y   CAR1d . However, regression

(1) is unable to show if subsequent performance of a …rm going public is a consequence of its

own underpricing or overall IPO market underpricing. Regression (2) tackles this issue by

including the average market underpricing in the month before the IPO. The estimates show

that the market e¤ect dominates as the …rm-speci…c underpricing no longer predicts future

long-term underperformance. Also, the magnitude of the long-term underperformance driven

by market underpricing is larger. A one percentage point increase in market underpricing

one month before the IPO implies a decline of 107bps in the 5y        CAR1d .

   Results in regression (2) shows that the long-term underperformance of IPO …rms is more

severe for IPOs in periods of high average market underpricing. Column (3) provides further

evidence. I replace the value of the average market underpricing by two dummies: Neutral

Month and High Month which are equal to one if the month preceding the IPO classi…es

as neutral- or high-underpricing, respectively. Firms going public after a high-underpricing

month perform substantial worse than …rms going public after a low-underpricing month.

   Regressions (4) and (5) show that the long-term underperformance in high-underpricing

periods is more pronounced for …rms going public after several high-underpricing months.

Including the number of consecutive low-, neutral- and high-underpricing months prior to

the IPO plus the interaction with …rm-speci…c underpricing, I …nd a negative relation be-

tween the number of consecutive high-underpricing months before the IPO and 5y         CAR1d .

One additional month in the number of consecutive high-underpricing months before the IPO

entails a decrease on 5y   CAR1d of 202 bps if I use the value of average market underpricing,

or, 454bps if I use dummies for average market underpricing. Firm-speci…c underpricing and

number of consecutive low-underpricing months do not seem to explain long-term underper-

formance.

   Table 4 presents evidence that rejects a relation between …rms going public in periods

of low/high volume and long-run performance. Regressing 5y             CAR1d on …rm speci…c

underpricing and number of IPOs in the month preceding the IPO, I …nd that high volume

                                               14
                     IPO underpricing and long-term underperformance


can predict subsequent poor long-run performance. However, adding the average underpricing

in the market in the month prior to the IPO shuts down both the …rm speci…c underpricing

and the IPO volume predictive power. Replacing number of IPOs by dummies representing

how "hot" the market is in terms of number of IPOs in the month prior to the IPO does not

change the results. Under the underpricing de…nition, long-term underperformance is more

severe for IPOs in later stages of high-underpricing periods. We do not observe the same

e¤ect using periods of high volume. Regressions (4) and (5) show that …rms going public

after several months of high volume do not experience worse long-run performance.

   Tables (3) and (4) provide evidence that, between …rm-speci…c underpricing, IPO volume,

and average market underpricing, the latter e¤ect is the dominant one in terms of predicting

long-run underperformance.

   Another way to address the relation between IPO underpricing and long-run underper-

formance is to compare wealth ratios between low, neutral and high underpricing/volume

periods. I compare how much money an investor gets by investing in an IPO …rm and hold-

ing it for …ve years to the same investment in the portfolio of …ve …rms matched on book

value, market capitalization, industry, and operating performance. A wealth ratio below one

means that the investor would have been better o¤ with an investment in the control …rms.

   Table 5 presents wealth ratios between IPO and benchmark …rms for the di¤erent phases

of the IPO cycle based on the underpricing and volume de…nitions. Panel 1 excludes …rst day

returns, which is the correct measure for an investor buying at the end of the …rst-trading day.

Panel 2 includes the …rst day return, thus, captures the return obtained by an investor who

is allocated shares at the IPO price. For the 1973-2008 period, investors who are allocated

shares at the IPO price do not lose money, as the wealth ratio of 0.97 is not statistically

di¤erent from one. However, an investor that buys at the end of the …rst trading date would

be better o¤ investing in the control …rms as she only achieves a wealth ratio of 0.84 by

investing in IPO …rms.

   Dividing the sample into periods of low-, neutral-, and high-underpricing periods, we

observe that long-term underperformance is more severe in periods of high-underpricing. Ex-

cluding the …rst day return, IPOs in high-underpricing periods display a …ve year wealth ratio

of only 0.66, while, for IPOs in low-underpricing periods, the wealth ratio is 0.98. Moreover,

within high-underpricing periods, we see that long-run underperformance is particularly se-

                                              15
                     IPO underpricing and long-term underperformance


vere for IPOs in the later stages. In the end of high-underpricing periods, IPO …rms earn a

…ve-year buy-and-hold return of -7.8%, while benchmark …rms earn 61.2%. Measuring long-

run performance relative to the IPO price reveals an interesting result. Firms going public

in low and neutral periods do not under or over-perform benchmark …rms but …rms going

public in high-underpricing periods do underperform. Subdividing these high-underpricing

periods, we see that in the beginning there is no underperformance but in later stages this

underperformance is large as evidenced by a wealth ratio of 0.71.

   This result is di¢ cult to reconcile with the view that IPO underpricing is a discount to

fundamental value. Assuming that the long-run price proxies the fundamental value, it seems

that IPO prices are set equal to fundamental value in low-, neutral-, and beginning of high-

underpricing periods; and above fundamental value in the late stages of high-underpricing

periods.

   Under a classi…cation of periods based on volume we do not see major di¤erences in long-

term underperformance. Excluding the …rst day return, there are no di¤erences between

neutral- and high-volume periods or within high-volume periods. Including the …rst day

return, all wealth ratios are not signi…cantly di¤erent from one.

   In this section I presented the two main …ndings of this paper. The …rst …nding is that

…rms going public in low-underpricing periods do not underperform in the long-run but …rms

that go public in high-underpricing do. This long-run underperformance is due to …rms going

public in the later stages of high-underpricing periods. The second …nding is that …rms that

issue in the late stages of high-underpricing periods exhibit long-term underperformance even

relative to the o¤er price.

   Established this relation between average market underpricing and long-term underper-

formance, in the next section, I test if risk can explain the relation.


4.2   Long-term underperformance and risk

One way to rationalize the results in long-term underperformance presented in the previous

section is through risk. In order for risk to explain the worse long-run performance of …rms

going public in high-underpricing periods, we should see safer …rms going public in these

periods. Furthermore, …rm risk should decrease within high-underpricing periods given that

we observe worse long-run performance for …rms going public in the late stages of high-

                                               16
                     IPO underpricing and long-term underperformance


underpricing periods.

   I test this hypothesis by looking at several measures of risk: earnings volatility, delisting

                                     s
rates, volatility of returns, and …rm’ beta over the …ve-year period after the IPO. Earnings

volatility is computed as the standard deviation of reported quarterly earnings; volatility of

                                                                  s
returns is the standard deviation of monthly returns; and the …rm’ beta is estimated from

the CAPM. Quarterly earnings are retrieved from Compustat, and delisting codes and daily

returns come from CRSP. For delisting rates, the dependent variable is one if the …rm is

delisted or zero if it stays on CRSP for the …ve-year period. If the patterns in returns are

caused by risk we should observe lower earnings volatility, delisting rates, return volatility

and betas for …rms going public in high-underpricing periods and even lower for IPOs in the

late stages of high-underpricing periods.

   Tables 6 to 10 reject this claim. Table 6 shows there is no relation between earnings

volatility and IPO underpricing. Firm-speci…c underpricing and average market underpricing

at the time of the IPO is not related to subsequent earnings volatility. In Table 7, I present

logit estimates for delisting rates. Speci…cations (3) and (5) show that IPOs following a

high-underpricing month are more likely do be delisted from CRSP than IPOs following

a low-underpricing month. This e¤ect is not ampli…ed with number of high-underpricing

months preceding the IPO.

   Results in volatility of returns on Table 8 point into the same direction as delisting rates.

We see that …rm-speci…c underpricing is positively correlated with return volatility. Including

average market underpricing we see that higher market underpricing is associated with higher

subsequent volatility. Using dummies for level of underpricing in the market does not change

the result that …rms issuing after a high-underpricing month exhibit higher return volatility.

Adding the number of high-underpricing months preceding the IPO magni…es the e¤ect. One

concern about this measure is that it does not separate between idiosyncratic and systematic

risk. Table 9 addresses the issue by presenting results for CAPM betas. Results on betas

are consistent with the ones on delisting rates and total volatility. Firms going public in

high-underpricing periods display higher betas. Speci…cations (4) and (5) also show that

betas are higher for …rms going public after several consecutive high-underpricing months.

   These results suggest that, if anything, …rms going public in high-underpricing periods

are more risky, and de…nitely not safer, than …rms going public in low-underpricing periods.

                                              17
                        IPO underpricing and long-term underperformance


   In table 10, I present estimates for the following regressions:


      Yi =     0   +   1 N eutrali   +   2 Highi   + "i

      Yi =     0   +   1 Lowi   +    2 N eutrali   +   3 M iddle   of Highi +   4 End   of Highi + "i


   where the dependent variables are earnings volatility, delisting rates, return volatility and

CAPM betas. Neutral, High, Middle of High and End of High are dummy variables equal to

one if the IPO is undertaken in a neutral-, high-, middle of high-, or end of high-underpricing

period, respectively.

   In panel 1, we see that, in the …ve years following the IPO, …rms going public in high-

underpricing periods display higher total return volatility and higher betas, and are more

likely to be delisted. There are no di¤erences in terms of earnings volatility. Panel 2 shows

that …rms going public in the beginning of high-underpricing periods are more likely to get

delisted, exhibit higher earnings volatility, return volatility and higher betas than …rms going

public in low and neutral-underpricing periods. We do not observe signi…cant di¤erences

within high-underpricing periods.

   Results presented in this section are not consistent with a risk-based explanation for the

relation between average market underpricing and subsequent long-run underperformance.

We do not observe safer …rms going public in high-underpricing periods or in the late stages

of high-underpricing periods. Results suggest the opposite as it seems that riskier …rms decide

to go public in high-underpricing periods.



4.3   Peso problem


In the previous section, I showed that the documented relation between average underpricing

in the market and subsequent long term-underperformance is not driven by risk. An alterna-

tive explanation is that such relation is caused by a peso problem. Under this explanation,

di¤erent returns in the long run are explained by di¤erent growth opportunities at the time

of the IPO. For such explanation to hold, we would need that the average …rm going public

in high-underpricing is worse than the average …rm going public in low-underpricing periods,

and also that the probability of providing huge returns is higher for high-underpricing pe-

                                                       18
                     IPO underpricing and long-term underperformance


riods. The long-term underperformance comes from investors paying more for worse …rms,

hoping to discover the next Microsoft.

   I test this explanation by analyzing if …rms going public in high-underpricing periods are

more likely to provide extremely high returns in the long run.

   Table 10 provides suggestive evidence that the relation between IPO underpricing and

long-term underperformance is not driven by a peso problem. Table 10 shows that IPOs

in high-underpricing periods are not more likely to provide extremely high returns, de…ned

as …ve-year buy-and-hold returns above 500 or 1000%. We actually observe that is more

likely to …nd the next Microsoft in …rms going public in low-underpricing periods than in

high-underpricing periods. Moreover, the probability of yielding extremely high returns is

the lowest in the late stages of high-underpricing periods, when long-term underperformance

is more severe.

   Table 11 reinforces results in Table 10. Performing logit regressions where the dependent

variable is one if the …rm provides extremely high returns, I do not …nd that IPOs in high-

underpricing environments are more likely to provide these huge returns in the long run.

Furthermore, looking at speci…cation (5), we see that …rms going public after several high-

underpricing months are less likely to yield extremely high returns after the IPO. This result is

inconsistent with the proposed explanation. If in fact, the relation between IPO underpricing

and long-term underperformance is caused by a peso problem, we should see exactly the

opposite.



4.4   IPO timing


In previous sections, I showed that …rms going public in the late stages of high-underpricing

periods display the worst subsequent long-run performance. These …rms even underperform

relative to the o¤er price. This result suggests that, during high-underpricing periods, IPO

prices are set above fundamental value. The idea that …rms are sold at a premium although

uncommon is not an original one. Using intrinsic value based on industry-matched price/sales

and price/EBITDA from comparable publicly traded …rm, Purnanandam and Swaminathan

(2001) …nd that o¤er prices are priced 50% above comparables. This premium over funda-

mental value can be the result of …rms trying to exploit excessively optimistic investors. If

                                               19
                    IPO underpricing and long-term underperformance


that is the case, then …rms have an incentive to time their IPOs in order to coincide with

periods of overvaluations by investors.

   Pagano et al. (1998) …nds support that …rms try to time their IPOs so they can take

advantage of industry-wide overvaluations, rather than to …nance their growth. Based on a

sample of Italian …rms, and using both ex ante and ex post information on characteristics

and performance, they …nd that …rms are more likely to go public when the average market-

to-book ratio of public …rms is high. Moreover, they …nd that higher market-to-book ratios

          ect
do not re‡ investment opportunities as companies tend do go public after and not before

periods of high investment.

   We see …rms going public in low-underpricing periods, when it seems there is no premium

to fundamental to be exploited. It might be that these …rms decide to go public because

they need capital to …nance investment opportunities. The optimal scenario is to go public

when IPO prices include a premium but if …rms cannot wait because, for example, investment

opportunities be exploited by others, then …rms are willing to undertake an IPO even in a

low-underpricing period. In high-underpricing periods, given the premium to fundamental

value, both …rms with and without investment opportunities have an incentive to go pub-

lic. Hence, post-IPO, we should observe di¤erences in …rms’ characteristics between …rms

going public in low- and high-underpricing periods. I test this claim by looking at opera-

tional performance measured as return on assets (ROA), total pro…tability, investment rates,

asset growth, and cash holdings. ROA is de…ned as operating income before depreciation

over assets; total pro…tability is income before extraordinary items over assets; investment

rate is capital expenditures over assets; and cash holding is cash over assets. All data is

retrieved from Compustat. Everything else the same, after the IPO, …rms going public in

low-underpricing periods should display higher ROA, higher pro…tability, should invest more,

hold less cash, and exhibit higher asset growth.

   I run the same set of regressions as in previous sections, where the dependent variables

are the di¤erences between the IPO …rm and control …rms, in each speci…c measure, and one

year after the IPO. Di¤erences in each measure are winsorized at the …ve percent level. Inde-

pendent variables are …rm-speci…c underpricing, average market underpricing in the month

prior to the IPO, dummies for neutral and high month if the month prior to the IPO clas-

si…es as neutral- or high-underpricing month, and number of consecutive low-, neutral-, or

                                             20
                       IPO underpricing and long-term underperformance


high-underpricing months prior to the IPO.

   Tables 13 to 18 present evidence that …rms going public in high-underpricing periods

seem to be of lower quality as they exhibit, after the IPO, worse ROA and pro…tability,

invest less, grow more slowly, and hold more cash. Moreover, these patterns are substantially

more pronounced for …rms going public in the late stages of high-underpricing periods.

   Table 13 and Table 14 present the results for ROA and overall pro…tability. Regressions

(1) and (2) show that …rm-speci…c and average market underpricing have a negative e¤ect

on ROA and pro…tability. These e¤ects are economically signi…cant as an increase of one

percentage point in average underpricing in the month prior to the IPO predicts a decrease

of around 17 basis points in ROA and pro…tability. Replacing average market underpricing

by dummies representing level of underpricing in the market still yields negative coe¢ cients

for high underpricing but breaks the statistical signi…cance of the e¤ect. I also …nd that the

number of consecutive high underpricing months has a negative impact in future ROA and

pro…tability. The fact that …rms going public in the late stages of high-underpricing periods

register the worst performance in terms of subsequent ROA is consistent with a story in which

low quality …rms rush into the market trying to exploit a premium to fundamental value.

   Results in asset growth and investment rates are presented in Tables 15 and 16. We see

that average market underpricing in the month prior to the IPO is negatively related with

asset growth and investment rates on the …rst year after the IPO. Moreover, speci…cation (5)

suggest, once again, that …rms going public after several high-underpricing months are the

ones experiencing the lowest asset growth and lowest investment rates.

   Finally, results concerning cash holdings are presented in Table 17. In this case, …rm-

speci…c underpricing, rather than market underpricing, has a positive e¤ect on cash holdings.

However, regressions (5) and (6) show that …rms going public after several high-underpricing

months hold more cash.

   An alternative way to test di¤erences in post-IPO performance is to run the following

regressions:


     Yi =      0   +   1 N eutrali   +   2 Highi   + "i

     Yi =      0   +   1 Lowi   +    2 N eutrali   +   3 M iddle   of Highi +   4 End   of Highi + "i



                                                       21
                     IPO underpricing and long-term underperformance


   where dependent variables are the performance measures presented in this subsection.

Neutral, High, Middle of High and End of High are dummy variables equal to one if the

IPO is undertaken in neutral-, high-, middle of high-, or end of high-underpricing period,

respectively.

   Panel 1 of Table 18 con…rms that …rms going public in high-underpricing periods exhibit,

post-IPO, lower ROA, pro…tability, asset growth, and higher cash holdings. Economic mag-

nitude of the e¤ects is large, ranging from six to seven percentage points. Panel 2 shows

that the di¤erences observed in Panel 1 are caused by …rms going public in the middle- and

end of high-underpricing periods. Firms going in the late stages of high-underpricing periods

display substantial lower ROA, pro…tability, asset growth, investment rates and higher cash

holdings. There are no signi…cant di¤erences between post-IPO performance of IPOs in the

beginning of high-underpricing periods and neutral- or low-underpricing periods.

   Results shown in this section are consistent with a story in which …rms try to time their

IPOs to coincide with periods of high underpricing, when IPO prices include a premium

to fundamental value. This explains why the number of IPOs goes up when underpricing

is high (recall Table 2). Firms with and without investment opportunities rush into the

market to take advantage of overvaluations. It is still puzzling why …rms do not fully exploit

overvaluations as they still leave money on the table, but is not a puzzle that they prefer to

go public when prices are above fundamental.


4.5   Underpricing periods and investor sentiment


In this section, I test if investor sentiment is stronger in high-underpricing periods. Results

presented in previous sections are consistent with a story in which …rms in high-underpricing

periods try to extract rents from overly optimistic investors in the aftermarket. For this to

hold, we should observe stronger investor sentiment in these periods. Clearly, this is not an

easy task since measuring sentiment is di¢ cult and no consensual measure exists. I propose to

address the issue by looking at the behavior on the …rst-trading day, correlation of sentiment

proxies with underpricing, and investors’response to earnings announcements.

   First, I look at the trading behavior on the …rst-trading day. If high-underpricing periods

are driven by sentiment investors we should observe more "excitement" on the …rst-trading


                                              22
                     IPO underpricing and long-term underperformance


day of high-underpricing periods. I test this claim by regressing the number of stock trades

and turnover percentage, measured as volume over number of shares outstanding, on the …rst-

trading day on dummies representing neutral-, and high-underpricing periods. I also analyze

sub periods of high-underpricing periods. In summary, I perform the following regressions:


               Yi =     0   +   1 N eutrali   +   2 Highi     + ui

               Yi =     0   +   1 Low   +   2 N eutrali   +    3 M iddlei   +   4 Endi   +   i



   where dependent variables are number of stock trades and turnover.

   Estimates on Table 19 show that in high-underpricing periods, number of stock trades,

and turnover on the …rst-trading day of …rms going public in high-underpricing periods is

higher than in low-underpricing periods. Panel 1 shows that, in low-underpricing periods,

16% of all shares outstanding of an IPO …rm are transacted on the …rst-trading day. This is

done with roughly 2,300 trades. Turnover for …rms going public in high-underpricing periods

is 27%. Furthermore, this eleven percentage point di¤erential in turnover is accompanied by

triple the number of trades. Panel 2 provides further insights on the question. Within high-

underpricing periods, I do not …nd signi…cant di¤erences in number of trades, and turnover

on the …rst-trading day. These results suggest that "excitement" on the …rst-trading day is

higher in high-underpricing periods and remains relatively constant throughout these periods.

   As a second approach to test if high underpricing is correlated with investor sentiment, I

analyze the relation between IPO underpricing and widely used proxies of investor sentiment.

Namely, I use the same variables as Baker in Wurgler (2006): dividend premium, number of

IPOs, NYSE turnover, closed-end fund discount, percentage of equity in total issues, and I

add the University of Michigan Consumer Con…dence Index (CCI).

   The dividend premium is de…ned as the di¤erence between the average market-to-book-

value ratios of dividend payers and nonpayers (Baker and Wurgler, 2004). Dividend-paying

stocks have a more predictable stream of incomes which is a characteristic of safety. Baker

and Wurgler (2004) claim that …rms when deciding whether to pay dividends appear to

accommodate sentiment for or against "safety". Higher dividend premium is interpreted as

a signal that investors are pessimistic and looking for safer investments.

   IPO volume is commonly associated with investor sentiment. Occasionally, there are


                                                   23
                       IPO underpricing and long-term underperformance


"windows of opportunity" that seem to ease the process of going public. If these "windows

of opportunity" are triggered by optimistic investors then we should see a positive relation

between number of IPOS and investor sentiment for a given period.

   The NYSE turnover is the ratio of trading volume to the number of shares listed on the

New York Stock Exchange. Volume can be viewed as an investor sentiment index. If short-

selling is more costly than opening and closing long positions, then sentiment investors are

more likely to trade when they are optimistic (Baker and Stein, 2004). Hence, high NYSE

turnover suggests high investor sentiment.

   The closed-end fund discount is another used proxy for investor sentiment. Closed-end

funds are funds with a …xed number of shares that trade on stock exchanges. The closed-end

fund discount (sometimes premium) is the di¤erence between the net asset value of the fund’s

                               s
security holdings and the fund’ market price. If closed-end funds are largely held by retail

investors, the observed discount can proxy sentiment (Zweig, 1973; Lee, Schleifer, and Thaler,

1991; Neal and Wheatley, 1998). When the average discount on closed-end funds goes up it

suggests that retail investors are pessimistic.

   The percentage of equity in total issues is also argued by Baker and Wurgler (2000) to

be related to investor sentiment. They …nd that the equity share can foretell low stock

market returns and that the pattern can be the result of …rms shifting between equity and

debt, successfully reducing cost of capital. A high equity share suggests high optimism by

sentiment investors.

   I also look at the University of Michigan Consumer Con…dence Index (CCI). Although this

survey does not ask directly consumers for their views on securities prices, changes in the CCI

correlate highly with changes with the UBS/Gallup index that targets investors. Moreover,

Qiu and Welch(2006), and Lemmon and Portniaguina (2006) show that CCI changes correlate

especially strongly with small stocks and returns of …rms held mainly by retail investors.

Hence, a positive relation between CCI and investor sentiment is expected.

                                            ect
   These investor sentiment proxies also re‡ economical fundamentals to some extent.

For example, IPO volume is partially determined by investment opportunities. To iron out

information about fundamentals I regress each proxy on a set of macroeconomic variables:

growth in industrial production, real growth in durable, nondurable, and services consump-

tion, growth in employment, and an NBER recession indicator. I use the residuals from these

                                                  24
                       IPO underpricing and long-term underperformance


regressions as sentiment proxies.

   Sentiment proxies are then dependent variables of the following regression:


                         Sentimenti =   0   +   1 N eutrali   +   2 Highi   + ui


   Table 20 shows that chas the predicted sign for the six sentiment proxies, although the
                        2

closed-end fund discount is not statistically signi…cant. The relation is particularly strong

for the dividend premium, percentage of equity in total issues and the CCI. For number of

IPOs and NYSE turnover, although we observe a signi…cant di¤erence between high- and

low-underpricing periods, we cannot reject that sentiment in high- and neutral-underpricing

periods is di¤erent.

   Finally, I test if in high-underpricing periods investors are being more optimistic than

usual. It might be the case that investors are not only optimistic about IPO …rms and

that they can also drive up other prices. One way to test if optimism is indeed higher in

high-underpricing periods is to look at price reactions to news. For example, price reactions

surrounding earnings announcements. The motivation is that earnings surprises generate

immediate reactions in prices. If in certain periods investors are more optimistic, then we

should observe weaker negative reactions to negative surprises and stronger positive reactions

to positive surprises.

   In order to test if price reactions to earnings announcements are more favorable in peri-

ods of high underpricing in the IPO market, I retrieve data from I/B/E/S, Compustat and

CRSP from January 1, 1984 to December 31, 2008. To determine the surprise in earnings

announcements I collect actual and forecasted earnings. I retrieve all quarterly earnings an-

nouncements from I/B/E/S for which there is at least one analyst forecast 90 days before the

announcement. During the 90 days prior to the earnings announcements we may observe sev-

eral updates on analyst forecasts. I only keep the forecast closest to earnings announcement

yielding a sample of 307,442 observations. It is important that we have the correct earnings

announcement dates. Hence, I also retrieve earnings announcement dates from Compustat.

The sample is restricted to observations for which there are announcements dates on I/B/E/S

and Compustat and di¤erence between the two dates is less than …ve days. In order to impute

a unique date to each earnings announcement I follow the optimal imputation rule used in


                                                 25
                      IPO underpricing and long-term underperformance


Dellavigna and Pollet (2009):

     when I/B/E/S and Compustat dates disagree I impute the earnings announcement date

to be the earliest;

     for the period before January 1, 1990 if the two dates agree I impute the announcement

date to be the previous trading day to the I/B/E/S and Compustat date;

     for the period after January 1, 1990 if the two dates agree I impute the announcement

date to be the I/B/E/S and Compustat date;

   This yields a sample size of 179,590 observations.

   I de…ne earnings surprise as the di¤erence between expected and actual earnings per share

normalized by share price as in Kotari (2000). For expected earnings I use I/B/E/S consensus

analyst forecast de…ned as the median forecast among all analysts. To eliminate any noise

in prices I use the price seven business days before the announcement. All observations for

which this information is not available and all penny stocks (price below $1) are dropped.

Sample is then reduced to 171,222 observations. Then I compute st;k ; the earnings surprise

for …rm k in quarter t as:
                                                         b
                                                 et;k et;k
                                        st;k =
                                                     pt;k

                                                                      b
   where et;k is the actual earnings per share of …rm k in quarter t, et;k the forecasted earnings

per share and pt;k the corresponding price seven business days before the announcement date.

I drop 124 observations for which the actual earnings are larger in absolute value than the

price of the share.

   Given that market reaction to earnings surprises may be non-linear I sort them into

quantiles as in DellaVigna and Pollet (2009). I create a total of 11 quantiles. Quantiles 1

to 5 concern negative surprises, quantile 6 has all announcements with no surprises while

quantiles 7 to 11 have all positive earnings surprises. The breakpoints for the quantiles are

computed year by year. Given that we have more positive surprises than negative surprises,

we do not have the same number of observations per bin.

   The goal is to obtain abnormal returns for the day of the announcement and the next

one. Thus, I match the sample with CRSP to obtain daily returns and I estimate betas for




                                                 26
                      IPO underpricing and long-term underperformance


each …rm-quarter observation from the regression:


                                         Ru ; k =    t;k   +    t;k Ru;m



where Ru ; k and Ru;m are respectively the daily return for …rm k and market at day u and

u2[      300;     45] with      beeing the announcement date. Given b ; we can compute buy

and hold abnormal return for the period [ + h; + H] for …rm k, quarter t as

                                                                 h                            i
                   h;H
                  Rt;k =      +H
                             j= +h (1    + Rj;k )    1 b t;k          +H
                                                                     j= +h (1   + Rj;m )     1


   Finally, I obtain from Compustat market capitalization and book to market for each

…rm-quarter observation. The …nal sample has 163,216 observations.

   I divide the sample in two subgroups: announcements during high-underpricing periods

and announcements during low- or neutral- underpricing periods.

   Figure 3 depicts preliminary evidence that investors react more favorably to news dur-

ing high-underpricing periods. For each quantile of earnings surprises, I compute average

CAR0;1 for the high- and non-high underpricing subsamples. As expected, for all quan-

tiles, cumulative abnormal returns are higher in the high-underpricing sample. This is es-

pecially true for positive news. CAR0;1 for surprises in bin eleven, announced in low-, and

neutral-underpricing periods, is 2.15%. For the same level of surprises, announcements in

high-underpricing periods generate a CAR0;1 of 3.16% - one percentage point di¤erence in a

two day window.

   To rigorously analyze the question I run the following regressions:


(1) CARit =           +   1 Highit   +   1 Earn_Surpit         +     2 Highit    Earn_Surpit + "it

(2) CARit =           +   1 Highit   +   1 Earn_Surpit         +     2 Highit    Earn_Surpit +

                  +   1 Controlsit   +    2 Controlsit         Earn_Surpit + "it

(3) CARit =           +   1 Highit   +   1 BotQit   +    2 T opQit    +    3 Highit      BotQit +      4 Highit     T opQit +

                  +   1 Controlsit   +    1 Controlsit         BotQit +       2 Controlsit        T opQit + "it

(4) CARit =           +   1 Highit   +   1 Bot2Qit   +       2 T op2Qit   +   3 Highit     Bot2Qit +      4 Highit     T op2Qit +

                  +   1 Controlsit   +    1 Controlsit         Bot2Qit +        2 Controlsit       T op2Qit + "it


                                                        27
                        IPO underpricing and long-term underperformance


    where Highit is a dummy variable equal to one when earnings are announced in a high-

underpricing period in the IPO market; Earn_Surpit represent earnings surprises; Controlsit

include decile of market capitalization, decile of book to market, year and month of the

announcement; BotQ=Bot2Q is a dummy variable equal to one if the surprise is in quantile

1/1&2 of earnings surprises; T opQ=T op2Q is a dummy variable equal to one if the earnings

surprise is in quantile 11/10&11.

    Results in Table 21 con…rm that investors react more favorably to earnings surprises in

high-underpricing periods. In the four speci…cations, we see that c is positive and signi…cant.
                                                                   1

Controlling for decile of market capitalization, decile of book to market, year and month of

the announcement does not eliminate this e¤ect. Moreover, speci…cations (3) and (4) show

that this reaction is stronger for surprises in the two top bins. CARs are …fty basis points

higher for announcements in high-underpricing periods and an extra …fty basis points for

the most positive surprises. This is a big e¤ect and strong evidence that high-underpricing

periods are correlated with high optimism.

    Results presented in this section suggest investor sentiment is particular strong in pe-

riods where the IPO market exhibits high underpricing. This is consistent with a story in

which …rms time their IPOs to periods when they can take advantage of these "exuberantly"

optimistic investors.



5    Discussion about money left on the table


Evidence presented in this paper suggests …rms try to exploit sentiment investors in high-

underpricing periods. IPO prices in the late stages of high-underpricing periods are set

above fundamental value. However, we still observe high …rst day returns for these …rms

going public. It is still a puzzle why …rms do not fully exploit sentiment investors. In this

                y
section, I brie‡ discuss four potential explanations for this puzzle.

    The Ljungqvist, Nanda and Singh (2006) model has …rms selling their shares to interme-

diaries, who then try to unload them in the aftermarket. Sentiment investors might show

up in the market, driving prices up. When these investors are expected, …rms extract rents

from them by demanding higher o¤er prices from underwriters. However, the fact that …rms

exploit overly optimistic investors does not imply zero underpricing. High underpricing can

                                              28
                    IPO underpricing and long-term underperformance


still occur as compensation given by …rms to intermediaries for bearing the risk of carrying

overvalued IPO shares in inventory. Although consistent with the results presented in this

paper, the story has two drawbacks: within high-underpricing periods, the likelihood that

sentiment investors show up in the aftermarket increases, which should lead to a decrease in

the level of underpricing within these periods. The data does show this decline in underpric-

ing. Moreover, the empirical magnitude of underpricing should provide strong incentives for

…rms to choose a di¤erent mechanism to go public such as auctions - we also do not observe

it in the data.

   A second explanation is based on information asymmetry. Unlike the traditional literature

on IPOs, the information asymmetry is not on fundamental value but on the maximum

value that sentiment investors are willing to pay. In this sense, the high …rst day return

is a discount not on fundamental value, but on sentiment investors’ valuations. I examine

analysts’earnings forecasts to see if periods of high underpricing are associated with a high

degree of information asymmetry. Table 22 shows that errors in earnings forecasts tend to be

higher in high-underpricing periods than in low underpricing. Also, the number of analysts’

forecasts per …rm tends to be smaller in high-underpricing periods. These two results are

consistent with a higher degree of information asymmetry in high-underpricing periods.

   A third possible explanation for why …rms do not fully exploit sentiment investors relies on

a very particular and strong form of sentiment –sentiment investors form their valuations by

anchoring on IPO prices. Imagine again that …rms sell shares to underwriters who then re-sell

them in the aftermarket. Assume that when sentiment investors show up in aftermarket they

always value shares at a premium over the IPO price. Under these assumptions, …rms and

underwriters cannot fully exploit sentiment investors. When sentiment investors are likely

to show up, …rms demand IPO prices at a premium to fundamental value, given the high

probability that underwriters can re-sell at a higher price to sentiment investors. However,

given the anchoring on IPO prices, there is no way for …rms to fully extract rents from

sentiment investors. Two problems of this explanation are the strong assumption on the

nature of investors’ sentiment, and the strong incentive to perform the IPO through an

auction.

   A fourth and perhaps the most reasonable explanation is that underwriters and …rms

share the rents from exploiting sentiment investors. Underwriters might be able to convince

                                             29
                        IPO underpricing and long-term underperformance


…rms to not fully exploit sentiment investors in return for better deals in the future. Firms

are willing to go public as they are sold above fundamental value. Underwriters can then

unload these shares, priced above fundamental value but below what sentiment investors are

willing to pay, to favored clients in return for future participation in less attractive IPOs or

quid pro quos4 .

        Determining the exact reason for why …rms are willing to leave money on the table is

beyond the scope of this paper. However, understanding that IPO prices include a premium

helps to understand why …rms decide to go public in high-underpricing periods. The level of

underpricing remains a puzzle, but the puzzle of increased IPO volume can be explained as

a reaction of …rms to the presence of sentiment investors.




6        Conclusion


This paper documents a close connection between the average underpricing in the market

and the long-term underperformance of IPOs. I show that the average underpricing in the

market determines the degree of subsequent long-run performance.

        The two central …ndings of the paper are that the well documented long-term underper-

formance of IPOs is driven by …rms going public in periods of high underpricing; and that

underpricing is best understood not as a discount to fundamental value, but as a discount to

in‡ated …rst day prices.

        I provide evidence that di¤erences in long term underperformance cannot be explained

by di¤erences in risk, and it is unlikely to be driven by a peso problem.

        I also …nd that …rms going public in later stages of high-underpricing periods display

worse operating performance and pro…tability, lower asset growth, lower investment rates

and higher cash holdings. These results are consistent with a story in which some …rms go

public to take advantage of overvalued equity.
    4
     “CSFB allocated shares of IPOs to more than 100 customers who, in return, funneled between 33 and
65 percent of their IPO pro…ts to CSFB. These customers typically ‡   ipped the stock on the day of the IPO,
often gaining tremendous pro…ts. They then transferred a share of their ‡  ipping pro…ts to CSFB by way of
excessively high brokerage commissions. . . The customers paid these commissions on uneconomic, limited-risk
trades in highly liquid, exchange-traded shares unrelated to the IPO shares— trades that they e¤ected for the
sole purpose of paying IPO ‡  ipping pro…ts back to CSFB.” (SEC News Release 2002-14).




                                                     30
                     IPO underpricing and long-term underperformance


   Finally, I …nd that investor sentiment is stronger in high-underpricing periods. The

presence of “exuberantly” optimistic investors in the market drives prices up. Firms and

underwriters exploit these optimistic investors, but the fact that underpricing still remains

high implies that they do not fully adjust IPO prices. Why …rms still leave money on the table

is still an unresolved puzzle. However, results presented shed light on why the number of IPOs

goes up when underpricing goes up. The increase in volume in high-underpricing periods is

puzzling if higher underpricing represents a large discount to fundamental value, as suggested

by the prior literature. However, the increase in number of IPOs in high-underpricing periods

makes sense if high underpricing coincides with IPO prices that are set above fundamental

value.



References

 [1] Aggarwal, Reena, and Patrick Conroy, 2000, Price discovery in initial public o¤erings and the

    role of the lead underwriter, Journal of Finance 55, 2903-2922


 [2] Allen, Franklin, and Gerald R. Faulhaber, 1989, Signaling by underpricing in the IPO market,

    Journal of Financial Economics 23, 303-324.


 [3] Alti, Aydogan, 2005, IPO market timing, Review of Financial Studies 18, 1105-1138.

 [4] Baron, David, 1982, A model of the demand for investment banking advising and distribution

    services for new issues, Journal of Finance 37, 955-976.


 [5] Beatty, Randolph, and Jay Ritter, 1986. Investment banking, reputation, and the underpricing

    of initial public o¤erings, Journal of Financial Economics 15, 213-232.


 [6] Benveniste, Lawrence M., and Paul A. Spindt, 1989, How investment bankers determine the o¤er

    price and allocation of new issues, Journal of Financial Economics 24, 343-362.


 [7] Benveniste, Lawrence M. and William J. Wilhelm, 1990, A comparative analysis of IPO proceeds

    under alternative regulatory environments, Journal of Financial Economics 28, 173-208.


 [8] Bernardo, Antonio E., and Ivo Welch, 2001, On the evolution of overcon…dence and entrepre-

    neurs, Journal of Economics and Management Strategy 10, 301-330.



                                                31
                      IPO underpricing and long-term underperformance


 [9] Booth, James R., and Lena Chua, 1996, Ownership dispersion, costly information, and IPO

     underpricing, Journal of Financial Economics 41, 291-310.


[10] Booth, James R. and Richrad L. Smith, 1986, Capital raising, underwriting, and the certi…cation

     hypothesis, Journal of Financial Economics 15, 261-280.


[11] Brav, Alon, and Paul A. Gompers, 1997, Myth or reality? The long-run underperformance of

     initial public o¤erings: Evidence from venture and nonventure capital-backed companies, Journal

     of Finance 52, 1791-1821.


[12] Brennan, Michael J., and J. Franks, 1997, Underpricing, ownership and control in initial public

     o¤erings of equity securities in the UK, Journal of Financial Economics 45, 391-413.


[13] Chemmanur, Thomas J., 1993, The pricing of initial public o¤ers: A dynamic model with infor-

     mation production, Journal of Finance 48, 285-304.


[14] Choe, H., R. Masulis, and V. Nanda, 1993, Common Stock O¤erings across the Business Cycle,

     Journal of Empirical Finance, 1, 1-29.


[15] Daniel, Kent, David Hirshleifer, and A. Subrahmanyam, 1998, Investor psychology and security

     market under- and over-reactions, Journal of Finance 53, 1839-1886.


[16] Dellavigna, Stefano and Joshua M. Pollet, 2009, Investor Inattention and Friday Earnings An-

     nouncements, Journal of Finance 64, 709-749.


[17] Ellis, Katrina, Roni Michaely, and Maureen O’Hara, 2000, When the underwriter is the market

     maker: An examination of trading in the IPO aftermarket, Journal of Finance 55, 1039- 1074.


[18] Field, Laura C., and Dennis P. Sheehan, 2001, Underpricing in IPOs: Control, monitoring, or

     liquidity? Working paper, Pennsylvania State University.


[19] Hanley, Kathleen Weiss, 1993, The underpricing of initial public o¤erings and the partial adjust-

     ment phenomenon, Journal of Financial Economics 34, 231-250.


[20] Heaton, J. B., 2002, Managerial optimism and corporate …nance, Financial Management.

[21] Ho¤man-Burchardi, 2001, Clustering of Initial Public O¤erings, Information Revelation and

     Underpricing, European Economic Review, 45, 353-383.


                                                 32
                      IPO underpricing and long-term underperformance


[22] Hughes, Patricia J., and Anjan V. Thakor, 1992, Litigation risk, intermediation, and the un-

     derpricing of initial public o¤erings, Review of Financial Studies 5, 709-742.


[23] Ibbotson, R. G., 1975, Price performance of common stock new issues, Journal of Financial

     Economics 2, 235-272.


[24] Ibbotson, Roger G., Jody L. Sindelar, and Jay R. Ritter, 1988, Initial public o¤erings, Journal

     of Applied Corporate Finance 1,37-45.


[25] Ibbotson, R. J. Sindelar, and J. Ritter, 1994, The Market’ Problem with the Pricing of Initial
                                                              s

     Public O¤erings, Journal of Applied Corporate Finance, 7, 66-74.


[26] Jain, B. and O. Kini, 1994, The Post-Issue Operating Performance of IPO Firms, Journal of

     Finance, 49, 1699-1726.


[27] Jegadeesh, Narasimhan, Mark Weinstein, and Ivo Welch, 1993, An empirical investigation of

     IPO returns and subsequent equity o¤erings, Journal of Financial Economics 34, 153-175.


[28] Korajczyk, R., D. Lucas, and R. McDonald, 1991, The E¤ect of Information Releases on the

     Pricing and Timing of Security Issues, Review of Financial Studies, 4, 685-708.


[29] Lie, Erik, 2001, Detecting Abnormal Operating Performance: Revisited, Financial Management,

     77-91.


[30] Ljungqvist, A., V. Nanda, and R. Singh, 2006, Hot Markets, Investor Sentiment, and IPO Pricing,

     Journal of Business 79, 1667-1703.


[31] Loughran, Tim, and Jay R. Ritter, 2002, Why don’ issuers get upset about leaving money on
                                                    t

     the table in IPOs? Review of Financial Studies 15, 413-443.


[32] Lowry, Michelle, 2003, Why does IPO volume ‡uctuate so much? Journal of Financial Economics

     67, 3-40.


[33] Lowry, Michelle, and G. William Schwert, 2002, IPO market cycles: Bubbles or sequential learn-

     ing? Journal of Finance 57, 3, 1171-1200.


[34] Mello, Antonio S., and John E. Parsons, 1998, Going public and the ownership structure of the

     …rm. Journal of Financial Economics 49, 79-109.


                                                  33
                      IPO underpricing and long-term underperformance


[35] Michaely, R. and W. Shaw, 1994, The Pricing of Initial Public O¤erings: Tests of Adverse-

     Selection and Signaling Theories, Review of Financial Studies, 7, 279-319.


[36] Mikkelson and Shah, 1997, Performance of Companies Around Initial Public O¤erings, Journal

     of Financial Economics, 44, 281-307.


[37] Miller, Edward M., 1977, Risk, uncertainty, and divergence of opinion, Journal of Finance 32,

     1151-1168.


[38] Muscarella, Chris J. and Michael R. Vetsuypens, 1989, A simple test of Baron’ model of IPO
                                                                                  s

     underpricing, Journal of Financial Economics 24, 125-136.


[39] Pagano, Marco, Fabio Panetta, and Luigi Zingales, 1998, Why do companies go public? An

     empirical analysis, Journal of Finance 53, 27-64.


[40] Purnanandam, Amiyatosh K., and B. Swaminathan, 2001, Are IPOs underpriced? Working

     paper, Cornell University.


[41] Rajan, R., and H. Servaes, 1997, Analyst Following of Initial Public O¤erings, Journal of Finance,

     52, 507-529.


[42] Ritter, Jay R., 1984, The "hot issue" market of 1980, Journal of Business 57, 215-240.

[43] Ritter, Jay R., 1991, The long run performance of initial public o¤erings, Journal of Finance 46,

     3-28.


[44] Rock, Kevin, 1986, Why new issues are underpriced, Journal of Financial Economics 15, 187-212.

[45] Sherman, Ann, 2000, IPOs and long-term relationships: An advantage of bookbuilding, Review

     of Financial Studies 13, 697-714.


[46] Sherman, Ann, and Sheridan Titman, 2002, Building the IPO order book: Underpricing and

     participation limits with costly information, Journal of Financial Economics.


[47] Smith, Cli¤ord W., 1986, Investment banking and the capital acquisition process, Journal of

     Financial Economics 15, 3-29.


[48] Spatt, Chester S., and Sanjay Srivastava, 1991, Preplay communication, participation restric-

     tions, and e¢ ciency in initial public o¤erings, Review of Financial Studies 4, 709-726.

                                                  34
                      IPO underpricing and long-term underperformance


[49] Stoughton, Neal M., and Josef Zechner, 1998, IPO-mechanisms, monitoring and ownership struc-

     ture, Journal of Financial Economics 49, 45-77.


[50] Stoughton, Neal M., Kit Pong Wong, and Josef Zechner, 2001, IPOs and product quality, Journal

     of Business 74, 375-408.


[51] Taylor, Shelley E., and Jonathan D. Brown, 1988, Illusion and well-being: A social psychological

     perspective on mental health, Psychological Bulletin 103, 193-210.


[52] Tinic, Seha M., 1988, Anatomy of initial public o¤erings of common stock, Journal of Finance

     43, 789-822.


[53] Teoh, S., I. Welch, and T. Wong, 1998, Earnings Management and the Long-Run Market Per-

     formance of Initial Public O¤erings, Journal of Finance, 53, 1935-1974.


[54] van Bommel, Jos, and Theo Vermaelen, 2001, Post-IPO capital expenditures and market feed-

     back, Journal of Banking and Finance.


[55] Welch, Ivo, 1992, Sequential sales, learning, and cascades, Journal of Finance 47, 695-732.

[56] Welch, Ivo, 1989, Seasoned O¤erings, Imitation Costs, and the Underpricing of Initial Public

     O¤erings, Journal of Finance, 44, 421-449.


[57] Welch, Ivo and Ritter, Jay, 2002, A Review of IPO Activity, Pricing and Allocations, Journal of

     Finance 57, 1795-1828.




                                                  35
                   IPO underpricing and long-term underperformance




Figure 1. Number of IPOs and monthly average …rst day return of U.S. IPOs
from 1973 to 2008. The sample is composed of U.S. IPOs completed between January 1,
1973 and December 31, 2008 as reported by Securities Data Company (SDC). Unit o¤ers,
spin-o¤s, closed-end funds, IPOs with an o¤er price below $1.00, Real Estate Investments
Trusts (REITS) and …nancial …rms as described by SDC are excluded from our sample.




                                          36
                    IPO underpricing and long-term underperformance




Figure 2. IPO level of underpricing and volume from 1973 to 2008. The sample
is composed of U.S. IPOs completed between January 1, 1973 and December 31, 2008 as
reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s, closed-end funds, IPOs
with an o¤er price below $1.00, Real Estate Investments Trusts (REITS) and …nancial …rms as
described by SDC are excluded from our sample. High underpricing/volume periods consist
of three or more consecutive high underpricing/volume months.Low underpricing/volume
periods are de…ned analogously and all other periods still not de…ned are considered neutral
underpricing/volume periods.




                                            37
                    IPO underpricing and long-term underperformance




Figure 3. Cumulative Abnormal Return (CAR) on the day of earnings announce-
ment as a function of level of underpricing in the IPO market. The sample is com-
posed by 163,261 observations. Each observation corresponds to quarterly announcements
in I/B/E/S from January 1984 to December 2008 for which there is information on the …rm
in Compustat and on the stock in CRSP. Earnings surprises are computed as the di¤erence
between median analysts forecast on I/B/E/S and reported earnings scaled by the price one
week prior to the announcement date. Cumulative abnormal returns are raw buy-and-hold
returns adjusted by the beta from market model at the day of announcement and the next
one. Quantiles 1 to 5 have all negative surprises; quantile 6 has all announcements with no
surprises and quantiles 7 to 11 concerns positive surprises. Quantiles are constructed year
by year. high-underpricing periods consist of three or more consecutive high underpricing
months.low-underpricing periods are de…ned analogously and all other periods still not de-
…ned are considered neutral-underpricing periods. Standard errors are White corrected and
clustered by day of announcement. ***, ** and * represent, respectively, 10%, 5% and 1%
signi…cance levels.




                                            38
                    IPO underpricing and long-term underperformance




 Cohort   Number   Average    Number of IPOs    Average     Agreggate   Average Money   Agreggate Money
  Year      of      1st day     with 1st day    Proceeds    Proceeds    Left on Table    Left on Table
           IPOs      return         Info
  1973      64      25.2%            5               3.7      235.2          1.9              9.7
  1974       7       1.9%            2               4.2       29.4          0.1              0.3
  1975       5       8.7%            4              35.3      176.3          0.6              2.6
  1976      32       2.7%           23               7.2      229.5          0.2              4.1
  1977      18      11.8%           13               6.3      112.8          0.5              6.4
  1978      30      21.5%           17               6.6      197.2          1.1             19.1
  1979      48      12.7%           34               6.6      318.3          1.0             32.8
  1980      104     27.3%           58               9.3      967.1          3.0             175.7
  1981      259      9.9%           159              8.9      2304.1         0.8             129.0
  1982      81      12.4%           51              11.7      950.3          2.0             102.4
  1983      425     13.4%           306             16.6      7050.9         2.1             645.6
  1984      183      4.5%           131              9.1      1657.0         0.4             47.4
  1985      181     10.0%           139             14.9      2688.5         0.9             123.8
  1986      358      8.8%           333             27.8      9936.0         2.4             791.7
  1987      275      8.9%           267             24.4      6703.3         2.7             733.7
  1988      102      8.8%           99              28.7      2927.1         1.6             161.0
  1989      106     17.1%           103             37.0      3923.2         17.9           1841.2
  1990      111     11.7%           108             27.2      3014.4         4.4             475.8
  1991      243     13.4%           241             38.3      9314.8         5.5            1331.1
  1992      319     10.6%           315             41.0     13066.6         3.9            1214.8
  1993      414     14.5%           408             41.9     17327.9         5.8            2348.1
  1994      349     10.2%           346             36.1     12615.4         4.0            1380.4
  1995      416     23.1%           412             48.9     20355.7         11.0           4511.7
  1996      599     16.8%           591             46.2     27659.3         8.0            4726.6
  1997      414     15.9%           410             53.1     21993.1         7.4            3016.9
  1998      232     23.3%           231             55.1     12784.1         11.9           2738.6
  1999      399     78.6%           397             97.0     38704.0         80.2           31826.0
  2000      317     56.7%           313             99.7     31608.5         67.0           20985.8
  2001      59      17.4%           59              114.7     6765.2         19.2           1130.5
  2002      57       8.7%           56              145.7     8303.7         13.2            739.2
  2003      52      12.5%           52              134.2     6979.1         12.6            655.2
  2004      147     11.8%           146             136.8    20105.7         22.0           3217.8
  2005      135      9.7%           132             159.1    21484.9         12.9           1696.4
  2006      141     10.5%           137             167.8    23654.8         18.2           2500.2
  2007      137     12.5%           137             157.9    21634.0         20.7           2841.0
  2008      23       4.9%           21              237.1     5453.1         26.8            562.5
 Total     6842      20%           6256             47.8      9394.0         14.8           2575.7

Table 1 IPO activity in the U.S. from 1973 to 2008. The sample is composed of U.S.
IPOs completed between January 1, 1973 and December 31, 2008 as reported by Securities
Data Company (SDC). Unit o¤ers, spin-o¤s, closed-end funds, IPOs with an o¤er price below
$1.00, Real Estate Investments Trusts (REITS) and …nancial …rms as described by SDC are
excluded from our sample. Values in million of dollars and not in‡ation adjusted. Money left
on the table is computed as …rst day return times number of shares issued.


                                               39
                    IPO underpricing and long-term underperformance




                                    A. Underpricing                B. Volume
                                Number 1st Day Return        Number 1st Day Return
    Panel 1. Division by months
    Low Month                     898         5.4%             442          15.5%
    Neutral Month                2335        12.4%            1,873         22.4%
    High Month                   1626        39.8%            2,544         19.5%
    Panel 2. Division by periods
    Low Period                    688        6.30%             412          15.40%
    Neutral Period               2,664       11.8%            1,946          19.4%
    High Period                  1,507       41.6%            2,501          21.7%
    Beginning of High Period      302        39.3%             706           24.1%
    Middle of High Period         596        46.3%            1,039          23.7%
    End of High Period            609        38.2%             756           16.7%

Table 2 Number of IPOs and monthly average …rst day return by phase of the
IPO cycle. The sample is composed of U.S. IPOs completed between January 1, 1970
and December 31, 2008 as reported by Securities Data Company (SDC). Unit o¤ers, spin-
o¤s, closed-end funds, IPOs with an o¤er price below $1.00, Real Estate Investments Trusts
(REITS) and …nancial …rms as described by SDC are excluded from our sample. High
underpricing/volume periods consist of three or more consecutive high underpricing/volume
months.Low underpricing/volume periods are de…ned analogously and all other periods still
not de…ned are considered neutral underpricing/volume periods. Beginning, Middle and End
of High Periods result from the division of high underpricing/volume periods into three equal
parts.




                                             40
                    IPO underpricing and long-term underperformance




                                    (1)        (2)       (3)        (4)       (5)
       Constant                   -17.144    -2.021    -2.926    -11.811    -4.379
                                 -3.14***     -0.31     -0.29      -1.28     -0.37
       Underpricing                -0.43     -0.125    0.992      -0.085     1.728
                                 -5.13***      -1.6      1.04      -0.52      1.4
       Market Underpricing                   -1.068               -0.495
                                            -6.27***               -1.42
       Neutral Month                                   -13.662              -22.287
                                                         -1.04                -1.29
       Under * Neutral                                  -1.578               -2.468
                                                          -1.6               -1.94*
       High Month                                      -27.341                 -3.7
                                                        -1.84*                -0.19
       Under * High                                     -1.347                -1.85
                                                          -1.4                -1.47
       Number Low Months                                           4.451      1.717
                                                                    0.49       0.17
       Under * Num Low                                            -0.115     -0.863
                                                                   -0.33      -1.36
       Number Neutral Months                                       2.799      4.871
                                                                    1.54      1.96*
       Under * Num Neutral                                         0.025      0.129
                                                                    0.35       1.37
       Number High Months                                           -4.3     -7.953
                                                                 -2.02**   -4.54***
       Under Num High                                             -0.007     -0.006
                                                                   -0.43      -0.31
       n                           4851      4851      4851        4851       4851
       R2                          0.003     0.007     0.005       0.009      0.011
Table 3 Long term underperformance of IPO …rms as a function of initial un-
derpricing. The sample is composed of U.S. IPOs completed between January 1, 1973 and
December 31, 2008 as reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s,
closed-end funds, IPOs with an o¤er price below $1.00, Real Estate Investments Trusts (RE-
ITS) and …nancial …rms as described by SDC are excluded from our sample. Dependent
variable is the …ve year cumulative abnormal return between IPO …rms and a portfolio of
…ve public …rms matched on book value, market value, industry and operating performance.
Underpricing denotes the …rst day return of the IPO …rm; Neutral and High Month are dum-
mies variables representing the level of underpricing at the time of the IPO; Num Low, Num
Neutral and Num High denote, respectively, number of consecutive low, neutral and high
underpricing months prior to each IPO. Standard errors are White corrected and clustered
by period. ***, ** and * represent, respectively, 10%, 5% and 1% signi…cance levels.




                                            41
                     IPO underpricing and long-term underperformance




                                      (1)         (2)        (3)        (4)        (5)
      Constant                      -1.414       8.151     -2.042      0.757    -17.421
                                     -0.13        0.73      -0.13      0.05       -1.03
      Underpricing                  -0.413      -0.125       0.19     -0.674     -0.066
                                   -5.08***      -1.61       0.43    -5.26***     -0.14
      Market Underpricing                       -1.025
                                               -6.01***
      Number of IPOs                -0.526       -0.36                -0.568
                                    -1.84*       -1.28                 -1.54
      Neutral Vol Month                                    -16.667               7.965
                                                             -0.95                0.39
      Under * Neutral Vol                                   -0.626              -0.712
                                                             -1.35               -1.52
      High Vol Month                                       -16.706               2.828
                                                             -0.97                0.14
      Under * High Vol                                      -0.695              -0.767
                                                             -1.54               -1.61
      Num Low Vol Months                                              0.135      0.554
                                                                       0.17      0.69
      Under * Num Low Vol                                             0.075      0.055
                                                                       1.64      1.09
      Num Neutral Vol Months                                         -2.063     -2.314
                                                                     -1.90*     -1.96*
      Under * Num Neutral Vol                                         0.087      0.096
                                                                     2.18**     2.48**
      Num High Vol Months                                             0.392     -0.502
                                                                       0.35      -0.46
      Under * Num High Vol                                            0.039      0.055
                                                                      1.73*     2.40**
      n                             4851           4851    4851       4851       4851
      R2                            0.004          0.008   0.004      0.007      0.007
Table 4 Long term underperformance of IPO …rms as a function of level of IPO
volume. The sample is composed of U.S. IPOs completed between January 1, 1973 and
December 31, 2008 as reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s,
closed-end funds, IPOs with an o¤er price below $1.00, Real Estate Investments Trusts (RE-
ITS) and …nancial …rms as described by SDC are excluded from our sample. Dependent
variable is the …ve year cumulative abnormal return between IPO …rms and a portfolio of …ve
public …rms matched on book value, market value, industry and operating performance. Un-
derpricing denotes the …rst day return of the IPO …rm; Neutral and High Month are dummies
variables representing the level of volume at the time of the IPO; Num Low, Num Neutral
and Num High denote, respectively, number of consecutive low, neutral and high volume
months prior to each IPO. Standard errors are White corrected and clustered by period. ***,
** and * represent, respectively, 10%, 5% and 1% signi…cance levels.




                                              42
                    IPO underpricing and long-term underperformance




                                A. Underpricing                            B. Volume
                     r5yIPO r5yBench WR             t-stat    r5yIPO    r5yBench WR         t-stat
 Panel 1. Excluding 1st day return
 Low                   66.0        68.9    0.98      -0.21     73.7        82.5     0.95      -0.63
 Neutral               45.2        59.4    0.91     -1.82*     31.2        61.5     0.81   -3.78***
 High                  10.5        68.0    0.66    -5.79***    36.3        61.8     0.84   -3.29***
 Beginning of High     53.0        84.6    0.83      -1.16     44.8        72.4     0.84    -2.41**
 Middle of High          7.7       66.5    0.65    -3.56***    30.4        54.7     0.84     -1.72*
 End of High            -7.8       61.2    0.57    -5.74***    36.4        61.5     0.84    -2.11**
 Total                 37.4        63.4    0.84    -4.68***    37.4        63.4     0.84   -4.68***
 Panel 2. Including 1st day return
 Low                   78.8        68.9    1.06        0.66    103.7       81.8     1.12    1.18
 Neutral               62.3        59.7    1.02        0.34     51.8       61.4     0.94    -0.89
 High                  41.5        68.0    0.84     -1.97**     55.7       62.3     0.96    -0.78
 Beginning of High     112.5       85.4    1.15       0.66      71.2       73.1     0.99    -0.13
 Middle of High        33.3        66.2    0.80      -1.65*     50.3       55.2     0.97    -0.32
 End of High           14.3        61.1    0.71    -2.96***     48.5       61.9     0.92    -0.96
 Total                 58.2        63.4    0.97       -0.85     58.2       63.6     0.97     -0.8

Table 5 Five year wealth relatives between IPO …rms and benchmark …rms. The
sample is composed of U.S. IPOs completed between January 1, 1973 and December 31,
2008 as reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s, closed-end funds,
IPOs with an o¤er price below $1.00, Real Estate Investments Trusts (REITS) and …nancial
…rms as described by SDC are excluded from our sample. In Panel 1 …ve year buy and hold
returns are computed excluding the …rst day return; Panel 2 reports results including the
…rst day return. High underpricing/volume periods consist of three or more consecutive high
underpricing/volume months.Low underpricing/volume periods are de…ned analogously and
all other periods still not de…ned are considered neutral underpricing/volume periods. Begin-
ning, Middle and End of High Periods result from the division of high underpricing/volume
periods into three equal parts. Standard errors are White corrected and clustered by period.
***, ** and * represent, respectively, 10%, 5% and 1% signi…cance levels.




                                             43
                    IPO underpricing and long-term underperformance




                                    (1)       (2)       (3)        (4)       (5)
       Constant                  22.253    23.138     14.397     25.516    13.867
                                 3.40***   2.96***   18.20***   2.77***   15.07***
       Underpricing               -0.062    -0.046    -0.001     -0.061    -0.004
                                   -1.57    -1.75*     -0.03       -1.3     -0.12
       Market Underpricing                  -0.058               -0.148
                                             -0.61                -1.04
       Neutral Month                                  13.292               17.647
                                                        0.97                 1.03
       Under * Neutral                                -0.157               -0.202
                                                       -1.22                -1.17
       High Month                                      6.056                5.912
                                                        0.92                 0.93
       Under * High                                    -0.05               -0.054
                                                        -0.9                -0.74
       Number Low Months                                         -3.351    0.576
                                                                  -1.04     1.32
       Under * Num Low                                            0.032     0.003
                                                                   1.02      0.13
       Number Neutral Months                                     -0.674    -2.034
                                                                   -1.3     -1.09
       Under * Num Neutral                                          0      0.019
                                                                   0.03      0.83
       Number High Months                                         0.503     0.228
                                                                   0.35       0.7
       Under Num High                                             0.002        0
                                                                   0.38      0.15
       n                           3316      3316      3316       3316      3316
       R2                           0         0          0           0         0
Table 6 Earnings volatility of IPO …rms as a function of initial underpricing. The
sample is composed of U.S. IPOs completed between January 1, 1973 and December 31, 2008
as reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s, closed-end funds, IPOs
with an o¤er price below $1.00, Real Estate Investments Trusts (REITS) and …nancial …rms
as described by SDC are excluded from our sample. The dependent variable is the standard
deviation of quarterly earnings announced during the …ve years after the IPO. Underpricing
denotes the …rst day return of the IPO …rm; Neutral and High Month are dummies variables
representing the level of underpricing at the time of the IPO; Num Low, Num Neutral and
Num High denote, respectively, number of consecutive low, neutral and high underpricing
months prior to each IPO. Standard errors are White corrected and clustered by period. ***,
** and * represent, respectively, 10%, 5% and 1% signi…cance levels.




                                            44
                     IPO underpricing and long-term underperformance




                                     (1)        (2)         (3)       (4)        (5)
      Constant                     1.017      0.989        0.605    1.018      0.701
                                  15.26***   12.05***    4.02***   10.26***   3.78***
      Underpricing                    0          0        -0.013       0       -0.011
                                    0.27       0.44      -2.40**     0.11       -1.62
      Market Underpricing                     0.002                 -0.004
                                               1.01                    -1
      Neutral Month                                       0.538                 0.315
                                                         3.01***                 1.42
      Under * Neutral                                     0.008                 0.006
                                                           1.41                  0.81
      High Month                                          0.665                  0.59
                                                         3.83***              2.72***
      Under * High                                        0.012                  0.01
                                                         2.26**                  1.42
      Number Low Months                                              -0.186    -0.093
                                                                    -2.39**     -1.02
      Under * Num Low                                                -0.007    -0.002
                                                                     -1.80*     -0.48
      Number Neutral Months                                           0.069     0.065
                                                                     2.39**   1.98**
      Under * Num Neutral                                               0       0.001
                                                                      0.06       0.99
      Number High Months                                              0.051    -0.005
                                                                      1.58      -0.25
      Under Num High                                                    0          0
                                                                      0.32       0.38
      n                             4859       4859       4859        4859      4859
      R2                             0          0         0.013       0.011     0.017
Table 7 Delisting rates of IPO …rms as a function of initial underpricing. The
sample is composed of U.S. IPOs completed between January 1, 1973 and December 31, 2008
as reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s, closed-end funds, IPOs
with an o¤er price below $1.00, Real Estate Investments Trusts (REITS) and …nancial …rms
as described by SDC are excluded from our sample. Logit regressions where the dependent
variable is one if the IPO …rm gets delisted from CRSP during the …ve years after the IPO are
presented. Underpricing denotes the …rst day return of the IPO …rm; Neutral and High Month
are dummies variables representing the level of underpricing at the time of the IPO; Num
Low, Num Neutral and Num High denote, respectively, number of consecutive low, neutral
and high underpricing months prior to each IPO. Standard errors are White corrected and
clustered by period. ***, ** and * represent, respectively, 10%, 5% and 1% signi…cance levels.




                                             45
                     IPO underpricing and long-term underperformance




                                   (1)        (2)        (3)        (4)        (5)
     Constant                    4.947       4.519      4.755     4.715        5.02
                                67.23***   63.44***   35.15***   57.54***   33.25***
     Underpricing                 0.01      0.001      -0.004     0.004      -0.002
                                11.64***      1.25      -0.56     2.45**      -0.18
     Market Underpricing                     0.03                  0.019
                                           11.07***              4.92***
     Neutral Month                                     0.014                  -0.068
                                                        0.09                   -0.35
     Under * Neutral                                   0.012                   0.009
                                                        1.51                    0.97
     High Month                                        0.738                  -0.144
                                                      3.56***                  -0.63
     Under * High                                      0.012                   0.01
                                                        1.51                    1.09
     Number Low Months                                             -0.187     -0.259
                                                                 -3.58***   -4.60***
     Under * Num Low                                               -0.007     -0.004
                                                                   -1.72*      -0.93
     Number Neutral Months                                         -0.078     -0.089
                                                                 -3.81***   -3.44***
     Under * Num Neutral                                              0           0
                                                                    0.49        0.23
     Number High Months                                             0.091      0.197
                                                                  3.08***    5.79***
     Under Num High                                                   0       -0.001
                                                                  2.63***   -3.04***
     n                            4635      4635        4635        4635       4635
     R2                           0.03      0.088       0.05       0.099      0.089

Table 8 Volatility of returns of IPO …rms as a function of initial underpricing.
The sample is composed of U.S. IPOs completed between January 1, 1973 and December
31, 2008 as reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s, closed-end
funds, IPOs with an o¤er price below $1.00, Real Estate Investments Trusts (REITS) and
…nancial …rms as described by SDC are excluded from our sample. The dependent variable is
the standard deviation of monthly returns during the …ve years after the IPO. Underpricing
denotes the …rst day return of the IPO …rm; Neutral and High Month are dummies variables
representing the level of underpricing at the time of the IPO; Num Low, Num Neutral and
Num High denote, respectively, number of consecutive low, neutral and high underpricing
months prior to each IPO. Standard errors are White corrected and clustered by period. ***,
** and * represent, respectively, 10%, 5% and 1% signi…cance levels.




                                            46
                     IPO underpricing and long-term underperformance




                                   (1)        (2)        (3)        (4)        (5)
     Constant                    0.802      0.775       0.803      0.768      0.773
                                62.25***   50.73***   33.70***   37.23***   30.54***
     Underpricing                0.005      0.004       0.009      0.005       0.01
                                14.68***   11.86***   5.76***     7.52***    7.07***
     Market Underpricing                    0.002                  0.001
                                           2.70***                  0.69
     Neutral Month                                     -0.044                 -0.009
                                                        -1.51                  -0.28
     Under * Neutral                                   -0.002                 -0.003
                                                        -1.38                 -1.93*
     High Month                                         0.034                 -0.017
                                                         0.94                  -0.42
     Under * High                                      -0.004                 -0.005
                                                      -2.93***              -3.27***
     Number Low Months                                             0.031       0.031
                                                                 2.36**       2.36**
     Under * Num Low                                                 0        -0.002
                                                                    0.1      -2.09**
     Number Neutral Months                                        -0.003      -0.002
                                                                   -0.72        -0.4
     Under * Num Neutral                                             0        -0.001
                                                                    1.22    -2.93***
     Number High Months                                            0.017       0.024
                                                                 2.63***     4.21***
     Under Num High                                                  0           0
                                                                 2.92***       1.90*
     n                           4219       4219       4219        4219        4219
     R2                          0.123      0.127      0.128      0.135       0.139

Table 9 Betas of IPO …rms as a function of initial underpricing. The sample
is composed of U.S. IPOs completed between January 1, 1973 and December 31, 2008 as
reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s, closed-end funds, IPOs
with an o¤er price below $1.00, Real Estate Investments Trusts (REITS) and …nancial …rms
as described by SDC are excluded from our sample. The dependent variable is the beta from
the CAPM estimated for the …ve year period after the IPO. Underpricing denotes the …rst
day return of the IPO …rm; Neutral and High Month are dummies variables representing the
level of underpricing at the time of the IPO; Num Low, Num Neutral and Num High denote,
respectively, number of consecutive low, neutral and high underpricing months prior to each
IPO. Standard errors are White corrected and clustered by period. ***, ** and * represent,
respectively, 10%, 5% and 1% signi…cance levels.




                                            47
                       IPO underpricing and long-term underperformance




                                  Earnings     Delisting    Volatility    Beta
                                  Volatility    Rate       of Returns
             Panel 1.
             Low                    14.95       0.564        4.761         0.851
                                   15.49***    3.64***      30.39***     38.57***
             Neutral                8.718       0.518        0.015        -0.028
                                     0.83      2.91***        0.09         -1.04
             High                   4.265        0.62        1.222         0.187
                                     0.79      3.42***      5.49***      4.32***
             n                       3323        4928         4702         4273
             R2                        0        0.007         0.052         0.03
             Panel 2.
             Beginning of High      14.428        1.211       5.663        1.034
                                   12.37***     6.68***     24.07***     15.82***
             Low                    0.523        -0.648      -0.902       -0.183
                                      0.35     -2.72***     -3.19***     -2.65***
             Neutral                  9.24       -0.129      -0.887       -0.211
                                      0.88        -0.64     -3.59***     -3.14***
             Middle of High          -0.61       -0.065       0.636        0.037
                                     -0.46        -0.25       1.72*         0.39
             End of High            12.363       -0.004       0.172       -0.028
                                      0.97        -0.02       0.52         -0.33
             n                       3323         4928        4702         4273
             R2                        0          0.007       0.055        0.031
Table 10 Earnings volatility, delisting rates, volatility of returns and …ve year
betas of IPO …rms as a function of underpricing level at the time of the IPO.
The sample is composed of U.S. IPOs completed between January 1, 1973 and December
31, 2008 as reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s, closed-end
funds, IPOs with an o¤er price below $1.00, Real Estate Investments Trusts (REITS) and
…nancial …rms as described by SDC are excluded from our sample. Panel 1 and panel 2
present, respectively, estimates of dummy regressions: Yi = 0 + 1 N eutrali + 2 Highi + "i
and Yi = 0 + 1 Lowi + 2 N eutrali + 3 M iddle of Highi + 4 End of Highi + "i where
dependent variables are earnings volatility, delisting rate, return volatility and beta for the
…ve year period after the IPO. high-underpricing periods consist of three or more consecutive
high underpricing months.low-underpricing periods are de…ned analogously and all other
periods still not de…ned are considered neutral-underpricing periods. Beginning, Middle and
End of High Periods result from the division of high-underpricing periods into three equal
parts. Standard errors are White corrected and clustered by period. ***, ** and * represent,
respectively, 10%, 5% and 1% signi…cance levels.




                                               48
                    IPO underpricing and long-term underperformance


                                        >1000%               >500%
                                   Number Fraction      Number Fraction
               Low                   9       1.28%        30     4.27%
               Neutral               33      1.22%        76     2.81%
               High                  14      0.92%        40     2.63%
               Beginning of High     5       1.64%        14     4.59%
               Middle of High        5       0.83%        16     2.66%
               End of High           4       0.65%        10     1.62%
Table 11 IPOs yielding extremely high returns as a function of underpricing
level at the time of the IPO. The sample is composed of U.S. IPOs completed between
January 1, 1973 and December 31, 2008 as reported by Securities Data Company (SDC).
Unit o¤ers, spin-o¤s, closed-end funds, IPOs with an o¤er price below $1.00, Real Estate
Investments Trusts (REITS) and …nancial …rms as described by SDC are excluded from our
sample. Extremely high returns are de…ned as …ve-year buy-and-hold returns above 500 or
1000%. Buy-and-hold returns are computed excluding the …rst day return. High-underpricing
periods consist of three or more consecutive high underpricing months.low-underpricing pe-
riods are de…ned analogously and all other periods still not de…ned are considered neutral-
underpricing periods. Beginning, Middle and End of High Periods result from the division of
high-underpricing periods into three equal parts.




                                            49
                     IPO underpricing and long-term underperformance




                                  (1)         (2)         (3)         (4)         (5)
   Constant                     -3.428      -3.302      -3.261      -3.476      -3.633
                              -34.29***   -24.67***   -16.17***   -21.20***   -13.75***
   Underpricing                 -0.003      -0.001      -0.001       0.002       0.018
                                 -1.47       -0.32       -0.2        1.25        1.03
   Market Underpricing                       -0.01                   0.001
                                             -1.53                    0.15
   Neutral Month                                       -0.429                   -0.051
                                                       -1.72*                    -0.15
   Under * Neutral                                                               -0.02
                                                                                  -1.1
   High Month                                           0.055                    0.651
                                                         0.2                    2.03**
   Under * High                                        -0.005                   -0.017
                                                        -0.76                    -0.98
   Number Low Months                                                0.223        0.264
                                                                   2.12**       2.21**
   Under * Num Low                                                 -0.008       -0.014
                                                                   -1.71*        -1.57
   Number Neutral Months                                           -0.042        0.003
                                                                    -0.86         0.06
   Under * Num Neutral                                  0.001       0.001        0.001
                                                        1.38         0.86         1.24
   Number High Months                                              -0.044       -0.106
                                                                    -0.68      -2.19**
   Under Num High                                                  -0.002       -0.002
                                                                  -2.75***     -2.05**
   n                             4859        4859       4859        4859         4859
   R2                           0.0015      0.0041     0.0065      0.0152       0.0236

Table 12 Probability of yielding extremely high returns as a function of initial
underpricing. The sample is composed of U.S. IPOs completed between January 1, 1973
and December 31, 2008 as reported by Securities Data Company (SDC). Unit o¤ers, spin-
o¤s, closed-end funds, IPOs with an o¤er price below $1.00, Real Estate Investments Trusts
(REITS) and …nancial …rms as described by SDC are excluded from our sample. Logit
regressions where the dependent variable is one if the IPO …rm provides extremely high
returns. Extremely high returns are de…ned as …ve-year buy-and-hold returns above 500%.
Buy-and-hold returns are computed excluding the …rst day return. Underpricing denotes the
…rst day return of the IPO …rm; Neutral and High Month are dummies variables representing
the level of underpricing at the time of the IPO; Num Low, Num Neutral and Num High
denote, respectively, number of consecutive low, neutral and high underpricing months prior
to each IPO. Standard errors are White corrected and clustered by period. ***, ** and *
represent, respectively, 10%, 5% and 1% signi…cance levels.




                                            50
                     IPO underpricing and long-term underperformance




                                      (1)        (2)       (3)        (4)       (5)
        Constant                     1.508      3.933     1.844      2.34      -0.05
                                   2.65***    6.21***      1.39    2.97***     -0.03
        Underpricing                -0.098     -0.048     0.003     -0.051    -0.058
                                  -9.27***   -3.73***      0.04    -2.55**     -0.77
        Market Underpricing                    -0.176               -0.099
                                             -5.55***              -2.16**
        Neutral Month                                     -0.492               -0.054
                                                           -0.33                -0.03
        Under * Neutral                                   -0.031                0.032
                                                           -0.38                 0.36
        High Month                                        -1.835                3.933
                                                           -1.01                1.92*
        Under * High                                      -0.102                -0.04
                                                           -1.43                -0.51
        Number Low Months                                           1.245       1.863
                                                                   2.34**     2.86***
        Under * Num Low                                             0.072       0.074
                                                                     1.38       1.36
        Number Neutral Months                                       0.433       0.702
                                                                   2.05**     3.08***
        Under * Num Neutral                                         0.007       0.003
                                                                     0.74       0.23
        Number High Months                                         -0.521      -1.313
                                                                   -1.66*    -5.02***
        Under Num High                                                0         0.003
                                                                     0.01       1.49
        n                           4449          4449    4449      4449        4449
        R2                          0.023         0.038   0.026     0.042       0.043

Table 13 Post IPO operational performance (ROA) as a function of initial un-
derpricing. The sample is composed of U.S. IPOs completed between January 1, 1973 and
December 31, 2008 as reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s,
closed-end funds, IPOs with an o¤er price below $1.00, Real Estate Investments Trusts (RE-
ITS) and …nancial …rms as described by SDC are excluded from our sample. Dependent
variable is the di¤erence in return on assets(ROA) one year after the IPO between IPO …rms
and a portfolio of …ve public …rms matched on book value, market value, industry and oper-
ating performance. Di¤erences in ROA are winsorized at the …ve percent level. Underpricing
denotes the …rst day return of the IPO …rm; Neutral and High Month are dummies variables
representing the level of underpricing at the time of the IPO; Num Low, Num Neutral and
Num High denote, respectively, number of consecutive low, neutral and high underpricing
months prior to each IPO. Standard errors are White corrected and clustered by period. ***,
** and * represent, respectively, 10%, 5% and 1% signi…cance levels.




                                             51
                      IPO underpricing and long-term underperformance




                                      (1)         (2)        (3)        (4)       (5)
       Constant                      -0.36       1.676     -0.623     0.389     -1.734
                                      -0.7     2.98***      -0.56      0.58      -1.26
       Underpricing                 -0.107      -0.064      0.001     -0.073    -0.045
                                  -10.24***   -4.92***       0.02   -3.73***     -0.65
       Market Underpricing                      -0.148                -0.078
                                              -5.01***               -1.98**
       Neutral Month                                        0.398                -0.009
                                                             0.31                 -0.01
       Under * Neutral                                     -0.056                -0.002
                                                            -0.77                 -0.02
       High Month                                          -1.044                 3.719
                                                            -0.65               2.00**
       Under * High                                        -0.109                 -0.07
                                                           -1.66*                 -0.96
       Number Low Months                                             0.523       1.093
                                                                     1.14        1.91*
       Under * Num Low                                               0.068       0.056
                                                                     1.44         1.08
       Number Neutral Months                                         0.484       0.733
                                                                    2.57**      3.44***
       Under * Num Neutral                                           0.005          0
                                                                     0.57         0.01
       Number High Months                                           -0.538         -1.2
                                                                    -1.87*     -4.74***
       Under Num High                                                0.001       0.004
                                                                     0.57       1.99**
       n                             4473          4473    4473      4473         4473
       R2                            0.03          0.041   0.033     0.045        0.046
Table 14 Post IPO pro…tability as a function of initial underpricing. The sample
is composed of U.S. IPOs completed between January 1, 1973 and December 31, 2008 as
reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s, closed-end funds, IPOs
with an o¤er price below $1.00, Real Estate Investments Trusts (REITS) and …nancial …rms
as described by SDC are excluded from our sample. Dependent variable is the di¤erence
in pro…tability (Income Before Extraordinary Items/Assets) one year after the IPO between
IPO …rms and a portfolio of …ve public …rms matched on book value, market value, industry
and operating performance. Di¤erences in pro…tability are winsorized at the …ve percent
level. Underpricing denotes the …rst day return of the IPO …rm; Neutral and High Month
are dummies variables representing the level of underpricing at the time of the IPO; Num
Low, Num Neutral and Num High denote, respectively, number of consecutive low, neutral
and high underpricing months prior to each IPO. Standard errors are White corrected and
clustered by period. ***, ** and * represent, respectively, 10%, 5% and 1% signi…cance levels.




                                              52
                    IPO underpricing and long-term underperformance




                                    (1)         (2)       (3)       (4)      (5)
       Constant                   11.707      14.704     13.17   12.415    10.872
                                 11.97***   12.52***   6.02***   8.73***   4.48***
       Underpricing                0.05        0.113    0.365      0.124    0.325
                                   1.36      3.14***    2.50**    2.42**    1.91*
       Market Underpricing                    -0.217              -0.124
                                            -4.46***                -1.6
       Neutral Month                                    -2.394              -1.295
                                                         -0.96               -0.46
       Under * Neutral                                  -0.157              -0.139
                                                         -0.88               -0.67
       High Month                                       -4.571              0.719
                                                         -1.54                0.2
       Under * High                                      -0.32              -0.256
                                                       -2.10**               -1.41
       Number Low Months                                           1.916     2.271
                                                                  2.03**    2.32**
       Under * Num Low                                             0.135     0.053
                                                                    1.56      0.56
       Number Neutral Months                                       0.287     0.575
                                                                    0.75      1.47
       Under * Num Neutral                                         0.029     0.018
                                                                    1.06      0.58
       Number High Months                                         -0.444    -1.102
                                                                    -0.8   -2.53**
       Under Num High                                             -0.003       0
                                                                   -0.61     0.04
       n                           4474       4474      4474       4474      4474
       R2                          0.001      0.007     0.006      0.01      0.01
Table 15 Post IPO asset growth as a function of initial underpricing. The sample
is composed of U.S. IPOs completed between January 1, 1973 and December 31, 2008 as
reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s, closed-end funds, IPOs
with an o¤er price below $1.00, Real Estate Investments Trusts (REITS) and …nancial …rms
as described by SDC are excluded from our sample. Dependent variable is the di¤erence in
asset growth one year after the IPO between IPO …rms and a portfolio of …ve public …rms
matched on book value, market value, industry and operating performance. Di¤erences in
asset growth are winsorized at the …ve percent level. Underpricing denotes the …rst day
return of the IPO …rm; Neutral and High Month are dummies variables representing the
level of underpricing at the time of the IPO; Num Low, Num Neutral and Num High denote,
respectively, number of consecutive low, neutral and high underpricing months prior to each
IPO. Standard errors are White corrected and clustered by period. ***, ** and * represent,
respectively, 10%, 5% and 1% signi…cance levels.




                                            53
                    IPO underpricing and long-term underperformance




                                    (1)         (2)      (3)        (4)      (5)
       Constant                   3.583        4.093    3.873      3.361    3.294
                                 14.91***   13.06***   6.65***   8.56***   5.24***
       Underpricing               0.007        0.017    0.017      0.031    0.026
                                   1.03      2.87***     0.65    3.15***     0.85
       Market Underpricing                    -0.037              -0.017
                                            -2.89***               -0.79
       Neutral Month                                     -0.52              -0.492
                                                         -0.76               -0.65
       Under * Neutral                                   0.031               0.031
                                                          0.89                0.76
       High Month                                       -0.819              0.015
                                                         -1.14               0.02
       Under * High                                     -0.012              -0.006
                                                         -0.45               -0.17
       Number Low Months                                           0.575     0.572
                                                                    1.48      1.35
       Under * Num Low                                            -0.009    -0.007
                                                                   -0.38     -0.26
       Number Neutral Months                                        0.2      0.266
                                                                    1.47     1.85*
       Under * Num Neutral                                         0.001    -0.004
                                                                    0.06      -0.4
       Number High Months                                         -0.051    -0.144
                                                                   -0.38    -1.86*
       Under Num High                                             -0.002    -0.001
                                                                 -2.25**     -1.29
       n                           4406       4406      4406       4406      4406
       R2                           0         0.003     0.003      0.006     0.007
Table 16 Post IPO investment rate as a function of initial underpricing. The
sample is composed of U.S. IPOs completed between January 1, 1973 and December 31, 2008
as reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s, closed-end funds, IPOs
with an o¤er price below $1.00, Real Estate Investments Trusts (REITS) and …nancial …rms
as described by SDC are excluded from our sample. Dependent variable is the di¤erence in
investment rate (Capital Expenditures/Assets) one year after the IPO between IPO …rms and
a portfolio of …ve public …rms matched on book value, market value, industry and operating
performance. Di¤erences in investment are winsorized at the …ve percent level. Underpricing
denotes the …rst day return of the IPO …rm; Neutral and High Month are dummies variables
representing the level of underpricing at the time of the IPO; Num Low, Num Neutral and
Num High denote, respectively, number of consecutive low, neutral and high underpricing
months prior to each IPO. Standard errors are White corrected and clustered by period. ***,
** and * represent, respectively, 10%, 5% and 1% signi…cance levels.




                                            54
                     IPO underpricing and long-term underperformance




                                    (1)        (2)        (3)         (4)       (5)
      Constant                    11.33      11.142     11.324      11.918    12.672
                                 16.61***   13.59***    6.84***   12.17***    6.57***
      Underpricing                0.165       0.161      0.506       0.211     0.541
                                 10.36***    9.67***    5.91***    7.72***    4.48***
      Market Underpricing                     0.014                 -0.062
                                               0.39                  -1.27
      Neutral Month                                      -2.935               -3.717
                                                          -1.51                -1.58
      Under * Neutral                                    -0.193               -0.203
                                                        -1.99**                -1.53
      High Month                                          1.812               -1.812
                                                           0.85                 -0.7
      Under * High                                       -0.372               -0.382
                                                       -4.27***              -3.09***
      Number Low Months                                             -1.067     -1.33
                                                                     -1.47    -1.67*
      Under * Num Low                                                0.089    -0.044
                                                                   2.68***      -0.8
      Number Neutral Months                                         -0.667    -0.269
                                                                  -2.90***     -1.05
      Under * Num Neutral                                            0.003    -0.018
                                                                      0.2      -1.23
      Number High Months                                             0.902     0.67
                                                                   2.40**      1.90*
      Under Num High                                                -0.008    -0.005
                                                                  -3.43***    -1.81*
      n                           4471        4471       4471        4471      4471
      R2                          0.041       0.041      0.05        0.047     0.054
Table 17 Post IPO cash holdings as a function of initial underpricing. The sample
is composed of U.S. IPOs completed between January 1, 1973 and December 31, 2008 as
reported by Securities Data Company (SDC). Unit o¤ers, spin-o¤s, closed-end funds, IPOs
with an o¤er price below $1.00, Real Estate Investments Trusts (REITS) and …nancial …rms
as described by SDC are excluded from our sample. Dependent variable is the di¤erence in
cash holdings (Cash/Assets) one year after the IPO between IPO …rms and a portfolio of
…ve public …rms matched on book value, market value, industry and operating performance.
Di¤erences in cash holdings are winsorized at the …ve percent level. Underpricing denotes the
…rst day return of the IPO …rm; Neutral and High Month are dummies variables representing
the level of underpricing at the time of the IPO; Num Low, Num Neutral and Num High
denote, respectively, number of consecutive low, neutral and high underpricing months prior
to each IPO. Standard errors are White corrected and clustered by period. ***, ** and *
represent, respectively, 10%, 5% and 1% signi…cance levels.




                                             55
                      IPO underpricing and long-term underperformance




                         ROA        Pro…t     Asset Growth   Investment   Cash Holdings
  Panel 1.
  Low                     1.501      -1.18          16.141      3.769         13.988
                          1.22       -1.19         6.33***    6.14***        7.79***
  Neutral                 0.058      0.869          -2.724      0.196         -2.405
                          0.04       0.77              -1        0.29          -1.24
  High                   -7.153     -6.419          -6.941     -0.616          6.155
                        -3.57***   -3.66***        -2.09**      -0.82        2.82***
  n                       4513       4537            4538       4470           4535
  R2                      0.01       0.02            0.002      0.001         0.012
  Panel 2.
  Beginning of High      -0.196     -3.814          13.308      3.73          16.096
                          -0.11    -2.20**         3.55***    4.33***        9.12***
  Low                     1.697      2.634           2.833      0.038         -2.108
                           0.77       1.32            0.62       0.04          -0.84
  Neutral                 1.755      3.504           0.109      0.234         -4.512
                           0.91      1.93*            0.03       0.26        -2.35**
  Middle of High         -7.417     -5.333          1.078       0.529          5.154
                        -2.10**      -1.61           0.22        0.5           1.79*
  End of High            -6.356     -4.238         -11.114     -1.938          5.062
                        -2.17**     -1.66*         -2.27**    -1.96**         2.02**
  n                       4513       4537            4538       4470           4535
  R2                       0.02       0.02           0.005      0.003         0.013

Table 18 Post IPO performance measures as a function of underpricing level at
the time of the IPO. The sample is composed of U.S. IPOs completed between January
1, 1973 and December 31, 2008 as reported by Securities Data Company (SDC). Unit o¤ers,
spin-o¤s, closed-end funds, IPOs with an o¤er price below $1.00, Real Estate Investments
Trusts (REITS) and …nancial …rms as described by SDC are excluded from our sample. Panel
1 and panel 2 present, respectively, estimates of dummy regressions: Yi = 0 + 1 N eutrali +
  2 Highi +"i and Yi = 0 + 1 Lowi + 2 N eutrali + 3 M iddle of Highi + 4 End of Highi +"i
where dependent variables are the di¤erence in ROA, pro…tability, asset growth, investment
and cash holdings one year after the IPO between IPO …rms and a portfolio of …ve public
…rms matched on book value, market value, industry and operating performance. Di¤erences
are winsorized at the …ve percent level.: high-underpricing periods consist of three or more
consecutive high underpricing months.low-underpricing periods are de…ned analogously and
all other periods still not de…ned are considered neutral-underpricing periods. Beginning,
Middle and End of High Periods result from the division of high-underpricing periods into
three equal parts. Standard errors are White corrected and clustered by period. ***, ** and
* represent, respectively, 10%, 5% and 1% signi…cance levels.




                                              56
                     IPO underpricing and long-term underperformance




                                          Turnover    Number of Trades
                    Panel 1.
                    Low                     0.16            2280.27
                                          21.91***          5.39***
                    Neutral                 0.01             -59.17
                                            1.38              -0.12
                    High                    0.11            4641.50
                                          6.89***           4.04***
                    n                       4544              3079
                    R2                     0.056              0.089
                    Panel 2.
                    Beginning of High        0.28           4425.68
                                           9.29***          3.27***
                    Low                     -0.11          -2145.41
                                          -3.71***            1.66*
                    Neutral                 -0.10          -2204.58
                                          -3.24***           -1.72*
                    Middle of High           0.02           4040.96
                                             0.43             1.77*
                    End of High             -0.03          2047.64
                                            -0.95             0.97
                    n                       4544              3079
                    R2                      0.061             0.104
Table 19 Trading measures on the …rst trading day of IPO …rms as a function
of underpricing level at the time of the IPO. The sample is composed of U.S. IPOs
completed between January 1, 1973 and December 31, 2008 as reported by Securities Data
Company (SDC). Unit o¤ers, spin-o¤s, closed-end funds, IPOs with an o¤er price below $1.00,
Real Estate Investments Trusts (REITS) and …nancial …rms as described by SDC are excluded
from our sample. Panel 1 and panel 2 present, respectively, estimates of dummy regressions:
Yi = 0 + 1 N eutrali + 2 Highi + "i and Yi = 0 + 1 Lowi + 2 N eutrali + 3 M iddle of
Highi + 4 End of Highi +"i where dependent variables are number of trades and percentage
of total shares transacted on the …rst trading day. high-underpricing periods consist of
three or more consecutive high underpricing months.low-underpricing periods are de…ned
analogously and all other periods still not de…ned are considered neutral-underpricing periods.
Beginning, Middle and End of High Periods result from the division of high-underpricing
periods into three equal parts. Standard errors are White corrected and clustered by period.
***, ** and * represent, respectively, 10%, 5% and 1% signi…cance levels.




                                              57
                                   Dividend   Number of      NYSE        Closed-End      % Equity on    Michigan Consumer
                                   Premium       IPOs       Turnover    Fund Discount    Total Issues    Con…dence Index
                        Low           5.42       -11.88       0.055           2.692         -0.028            -1.422
                                    3.98***    -8.04***       1.73*         4.78***          -1.48            -1.65*
                        Neutral      -12.54       18.80       0.101          -4.388         -0.009             1.391
                                   -8.41***     8.13***      2.51**        -6.28***          -0.39             1.26
                        High         -18.94       18.49       0.091          -0.294         0.063              2.765
                                   -8.90***     6.99***       1.91*           -0.41        2.00**              1.73*




58
                        n              413         413         413             413            413               360
                        R2            0.245       0.131       0.016          0.125           0.021             0.009
     Table 20 Sentiment measures a function of underpricing level.Estimates on dummy regression: Yi = 0 + 1 N eutrali + 2 Highi + "i
     where dependent variables are monthly dividend premium, number of IPOs, NYSE trading volume, closed-end fund discount, percentage of
     equity on total issues and the University of Michigan Consumer Con…dence Index. Sample period is 1973 to 2008. high-underpricing periods
     consist of three or more consecutive high underpricing months.low-underpricing periods are de…ned analogously and all other periods still not
     de…ned are considered neutral-underpricing periods. Standard errors are White corrected and clustered by period. ***, ** and * represent,
     respectively, 10%, 5% and 1% signi…cance levels.
                                                                                                                                                     IPO underpricing and long-term underperformance
                     IPO underpricing and long-term underperformance




                                                     (1)         (2)      (3)         (4)
 Intercept                                         0.0010     0.0025    0.0035      0.0043
                                                   3.13***      1.13      1.59       1.94*
 High Underpricing Period                          0.0049     0.0049    0.0047      0.0048
                                                   6.16***    4.12***   3.95***     3.96***
 Earnings Surprise                                 0.1801     0.2159
                                                  11.12***      1.39
 High Underpricing * Earnings Surprise           -0.0047048    0.0016
                                                    -0.15       0.03
 Bottom Quantile                                                          -0.0217
                                                                        -21.12***
 Top Quantile                                                              0.0210
                                                                         26.08***
 High Underpricing * Bottom Quantile                                      -0.0035
                                                                            -1.62
 High Underpricing * Top Quantile                                          0.0054
                                                                         2.92***
 Bottom Two Quantiles                                                                 -0.0199
                                                                                    -27.96***
 Top Two Quantiles                                                                     0.0186
                                                                                     34.85***
 High Underpricing * Bottom Two Quantiles                                             -0.0025
                                                                                       -1.75*
 High Underpricing * Top Two Quantiles                                                 0.0025
                                                                                       1.97**
 Controls                                           NO         YES       YES            YES
 n                                                163,175     163,175   163,175       163,175
 R2                                                0.006       0.015     0.024          0.035

Table 21. Cumulative Abnormal Return (CAR) on the day of earnings announce-
ment as a function of level of underpricing in the IPO market. The sample is com-
posed by 163,261 observations. Each observation corresponds to quarterly announcements
in I/B/E/S from January 1984 to December 2008 for which there is information on the …rm
in Compustat and on the stock in CRSP. Earnings surprises are computed as the di¤erence
between median analysts forecast on I/B/E/S and reported earnings scaled by the price one
week prior to the announcement date. Cumulative abnormal returns are raw buy-and-hold
returns adjusted by the beta from market model at the day of announcement and the next
one. Quantiles 1 to 5 have all negative surprises; quantile 6 has all announcements with
no surprises and quantiles 7 to 11 concerns positive surprises. Quantiles are constructed
year by year.high-underpricing periods consist of three or more consecutive high underpric-
ing months.low-underpricing periods are de…ned analogously and all other periods still not
de…ned are considered neutral-underpricing periods. Standard errors are White corrected
and clustered by date of announcement. ***, ** and * represent, respectively, 10%, 5% and
1% signi…cance levels.




                                            59
                    IPO underpricing and long-term underperformance


                             Absolute Earnings     Number of Earnings
                                 Surprise              Forecasts
                   Low             0.005                   5.976
                                 14.56***               29.61***
                   Neutral         0.001                  -0.533
                                   2.21**                 -2.1**
                   High            0.001                  -0.904
                                   1.97**               -3.48***
                   n              163,220                163,220
                   R2              0.001                   0.004
Table 22. Information asymmetry as a function of level of underpricing in the IPO
market. The sample is composed by 163,261 observations. Each observation corresponds to
quarterly announcements in I/B/E/S from January 1984 to December 2008 for which there
is information on the …rm in Compustat and on the stock in CRSP. Estimates on dummy
regression: Yi = 0 + 1 N eutrali + 2 Highi + "i where dependent variables are the absolute
value of earnings surprise and number of forecasts. Earnings surprises are computed as the
di¤erence between median analysts forecast on I/B/E/S and reported earnings scaled by the
price one week prior to the announcement date. .high-underpricing periods consist of three or
more consecutive high underpricing months.low-underpricing periods are de…ned analogously
and all other periods still not de…ned are considered neutral-underpricing periods. Standard
errors are White corrected and clustered by period. ***, ** and * represent, respectively,
10%, 5% and 1% signi…cance levels.




                                             60

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:48
posted:9/26/2012
language:Unknown
pages:60